CursorBench 4.0
Evaluates coding agents on ambiguous multi-file tasks drawn from real Cursor sessions (v4.0).
- Category
- Agentic coding
- Unit
- Percent (%)
- Higher is better
- Comparison groups
- 3total
- Currently available model rankings · Ranked results: 20
- All models (records): Groups: 4 · Entries: 47
Rankings
Values with different source types (official, third-party, paper) or measurement conditions cannot be compared, so rankings are separated by comparison group (the same source, or sources verified to use the same conditions).
Rankings include only models currently available in Claude Code / Codex, excluding legacy models.
- View on rankings page (by metric)
- Cost balance (frontier and effort changes)
- Overall ranking (performance first)
official results (Introducing Claude Opus 5.5)
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Max | 57.8% | $13.43Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Max: 57.8% $13.43" |
| 2 | Claude Opus 5.5 | High | 56.0% | $3.97Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · High: 56.0% $3.97" |
| 2 | Claude Opus 5.5 | Xhigh | 56.0% | $6.99Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Xhigh: 56.0% $6.99" |
| 4 | Claude Opus 5.5 | Med | 52.5% | $2.90Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Med: 52.5% $2.90" |
| 5 | Claude Opus 5.5 | Low | 43.7% | $1.18Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Low: 43.7% $1.18" |
official results (Introducing Claude Sonnet 5.5)
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Max | 57.8% | $13.43Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Max: 57.8% $13.43" |
| 2 | Claude Opus 5.5 | High | 56.0% | $3.97Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · High: 56.0% $3.97" |
| 2 | Claude Opus 5.5 | Xhigh | 56.0% | $6.98Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Xhigh: 56.0% $6.98" |
| 4 | Claude Sonnet 5.5 | Max | 55.5% | $9.67Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Max: 55.5% $9.67" |
| 5 | Claude Sonnet 5.5 | Xhigh | 53.1% | $3.88Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Xhigh: 53.1% $3.88" |
| 6 | Claude Opus 5.5 | Med | 52.5% | $2.91Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Med: 52.5% $2.91" |
| 7 | Claude Sonnet 5.5 | High | 47.8% | $1.67Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · High: 47.8% $1.67" |
| 8 | Claude Opus 5.5 | Low | 43.7% | $1.17Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Low: 43.7% $1.17" |
| 9 | Claude Sonnet 5.5 | Med | 39.2% | $0.70Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Med: 39.2% $0.70" |
| 10 | Claude Sonnet 5.5 | Low | 35.8% | $0.50Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Low: 35.8% $0.50" |
official results (Claude Fable 5.2 and Claude Mythos 5.2 System Card)
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.2 | Max | 60.0% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt60.0%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 2 | Claude Fable 5.2 | High | 56.9% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt56.9%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 3 | Claude Fable 5.2 | Xhigh | 56.3% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt56.3%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 4 | Claude Fable 5.2 | Med | 54.8% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt54.8%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 5 | Claude Fable 5.2 | Low | 53.2% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt53.2%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
All model rankings (records)
These rankings are recalculated by comparison group including legacy models and models unavailable in Claude Code / Codex. They differ from the rankings above because the set of models differs.
- View all model rankings (by metric)
- Cost balance (all models, records)
- Overall ranking (all models, records)
Show rankings including legacy and unavailable models (entries: 47)
official results (Introducing Claude Opus 5.5)
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Max | 57.8% | $13.43Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Max: 57.8% $13.43" |
| 2 | Claude Opus 5.5 | High | 56.0% | $3.97Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · High: 56.0% $3.97" |
| 2 | Claude Opus 5.5 | Xhigh | 56.0% | $6.99Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Xhigh: 56.0% $6.99" |
| 4 | Claude Opus 5.5 | Med | 52.5% | $2.90Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Med: 52.5% $2.90" |
| 5 | Claude Fable 5.1 | Max | 51.8% | $17.28Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Fable 5.1 · Max: 51.8% $17.28" |
| 6 | Claude Fable 5.1 | Xhigh | 51.6% | $13.01Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Fable 5.1 · Xhigh: 51.6% $13.01" |
| 7 | Claude Fable 5.1 | High | 49.2% | $9.08Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Fable 5.1 · High: 49.2% $9.08" |
| 8 | Claude Fable 5.1 | Med | 46.8% | $7.05Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Fable 5.1 · Med: 46.8% $7.05" |
| 9 | Claude Opus 5 | Max | 46.6% | $11.95Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5 · Max: 46.6% $11.95" |
| 10 | Claude Opus 5 | Xhigh | 46.1% | $11.43Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5 · Xhigh: 46.1% $11.43" |
| 11 | Claude Fable 5.1 | Low | 45.1% | $5.44Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Fable 5.1 · Low: 45.1% $5.44" |
| 12 | Claude Opus 5 | High | 44.7% | $9.00Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5 · High: 44.7% $9" |
| 13 | Claude Opus 5.5 | Low | 43.7% | $1.18Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5.5 · Low: 43.7% $1.18" |
| 14 | Claude Opus 5 | Med | 43.3% | $6.94Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5 · Med: 43.3% $6.94" |
| 15 | GPT-5.6 Sol | Max | 41.7% | $8.23Cost per task | not stated | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Max" was mapped to max in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs Cost: "GPT-5.6 Sol · Max: 41.7% $8.23" |
| 16 | Claude Opus 5 | Low | 40.7% | $4.87Cost per task | Anthropic | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs Cost: "Opus 5 · Low: 40.7% $4.87" |
| 17 | GPT-5.6 Sol | Xhigh | 37.7% | $4.40Cost per task | not stated | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Xhigh" was mapped to xhigh in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs Cost: "GPT-5.6 Sol · Xhigh: 37.7% $4.40" |
| 18 | GPT-5.6 Sol | High | 35.7% | $2.85Cost per task | not stated | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "High" was mapped to high in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs Cost: "GPT-5.6 Sol · High: 35.7% $2.85" |
| 19 | GPT-5.6 Sol | Med | 31.1% | $1.77Cost per task | not stated | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Med" was mapped to medium in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs Cost: "GPT-5.6 Sol · Med: 31.1% $1.77" |
| 20 | GPT-5.6 Sol | Low | 24.6% | $0.87Cost per task | not stated | Introducing Claude Opus 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs Cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Low" was mapped to low in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs Cost: "GPT-5.6 Sol · Low: 24.6% $0.87" |
official results (Introducing Claude Sonnet 5.5)
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | Max | 57.8% | $13.43Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Max: 57.8% $13.43" |
| 2 | Claude Opus 5.5 | High | 56.0% | $3.97Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · High: 56.0% $3.97" |
| 2 | Claude Opus 5.5 | Xhigh | 56.0% | $6.98Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Xhigh: 56.0% $6.98" |
| 4 | Claude Sonnet 5.5 | Max | 55.5% | $9.67Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Max: 55.5% $9.67" |
| 5 | Claude Sonnet 5.5 | Xhigh | 53.1% | $3.88Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Xhigh: 53.1% $3.88" |
| 6 | Claude Opus 5.5 | Med | 52.5% | $2.91Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Med: 52.5% $2.91" |
| 7 | Claude Sonnet 5.5 | High | 47.8% | $1.67Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · High: 47.8% $1.67" |
| 8 | Claude Opus 5.5 | Low | 43.7% | $1.17Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Opus 5.5 · Low: 43.7% $1.17" |
| 9 | GPT-5.6 Sol | Max | 41.7% | $8.23Cost per task | not stated | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Max" was mapped to max in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs. cost: "GPT-5.6 Sol · Max: 41.7% $8.23" |
| 10 | Claude Sonnet 5.5 | Med | 39.2% | $0.70Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Med: 39.2% $0.70" |
| 11 | GPT-5.6 Sol | Xhigh | 37.7% | $4.40Cost per task | not stated | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Xhigh" was mapped to xhigh in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs. cost: "GPT-5.6 Sol · Xhigh: 37.7% $4.40" |
| 12 | Claude Sonnet 5.5 | Low | 35.8% | $0.50Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5.5 · Low: 35.8% $0.50" |
| 13 | GPT-5.6 Sol | High | 35.7% | $2.85Cost per task | not stated | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "High" was mapped to high in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs. cost: "GPT-5.6 Sol · High: 35.7% $2.85" |
| 14 | Claude Sonnet 5 | Max | 34.1% | $7.17Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5 · Max: 34.1% $7.17" |
| 15 | Claude Sonnet 5 | Xhigh | 32.0% | $4.55Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5 · Xhigh: 32.0% $4.55" |
| 16 | GPT-5.6 Sol | Med | 31.1% | $1.77Cost per task | not stated | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Med" was mapped to medium in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs. cost: "GPT-5.6 Sol · Med: 31.1% $1.77" |
| 17 | Claude Sonnet 5 | High | 30.8% | $3.48Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5 · High: 30.8% $3.48" |
| 18 | Claude Sonnet 5 | Med | 28.0% | $2.31Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5 · Med: 28.0% $2.31" |
| 19 | GPT-5.6 Sol | Low | 24.6% | $0.87Cost per task | not stated | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for a model from some other vendor (announced by Anthropic); the figure's effort label "Low" was mapped to low in the OpenAI API reasoning effortSource excerptCursorBench 4.0 — Accuracy vs. cost: "GPT-5.6 Sol · Low: 24.6% $0.87" |
| 20 | Claude Sonnet 5 | Low | 24.1% | $1.39Cost per task | Anthropic | Introducing Claude Sonnet 5.5Retrieved 2026-11-12 · Vendor-reported value. A point in the figure "CursorBench 4.0 / Accuracy vs. cost" on the Anthropic announcement page. values were taken from each point's aria-label (DOM); the page shows them only in the figure; value for its own model (announced by Anthropic)Source excerptCursorBench 4.0 — Accuracy vs. cost: "Sonnet 5 · Low: 24.1% $1.39" |
official results (Grok 4.7 (xAI news))
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Grok 4.7 | Xhigh | 46.3% | — | xAI | Grok 4.7 (xAI news)Retrieved 2026-11-12 · Vendor-reported value. Self-reported by xAI. Table heading is Grok 4.7 (xHigh)Source excerptCursorBench 4.0 | Grok 4.7 (xHigh) 46.3 |
| 2 | Grok 4.6 | High | 40.4% | — | xAI | Grok 4.7 (xAI news)Retrieved 2026-11-12 · Vendor-reported value. Self-reported by xAI (comparison column on the Grok 4.7 announcement page)Source excerptCursorBench 4.0 | Grok 4.6 (High) 40.4 |
official results (Claude Fable 5.2 and Claude Mythos 5.2 System Card)
| Rank | Model | effort | Score | Source cost (metric) | measured/reported by | source |
|---|---|---|---|---|---|---|
| 1 | Claude Fable 5.2 | Max | 60.0% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt60.0%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 2 | Claude Fable 5.2 | High | 56.9% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt56.9%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 3 | Claude Fable 5.2 | Xhigh | 56.3% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt56.3%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 4 | Claude Fable 5.2 | Med | 54.8% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt54.8%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |
| 5 | Claude Fable 5.2 | Low | 53.2% | — | Cursor | Claude Fable 5.2 and Claude Mythos 5.2 System CardRetrieved 2026-11-14 · Vendor-reported value. Independently run and reported by Cursor in its production agent harness. A single result per effort; no uncertainty interval. Additional prompt and harness improvements prevent direct comparison with values in the earlier Opus 5.5 and Sonnet 5.5 system cards. Numeric cost not stated in the body text.Source excerpt53.2%Location §8.8 “CursorBench 4.0”, body text, Fable 5.2 scores by effort |