Across 21 comparable benchmarks, Claude Sonnet 5 has better reported results on 12 and worse on 9. 1 further shared result is listed below, each with the reason it is not counted.
Moderate evidence21 of 22 shared results countedHow we compare
Price basis: Standard rates. The price section below shows the applicable periods.
| Fact | Mistral Large 4 | Claude Sonnet 5 |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | $0.700 | $2.00 |
| Output / 1M | $2.10 | $10.00 |
| Blended (4:1) | $0.526 | $2.30 |
| Status | Preview | Available |
| Released | October 6, 2026 | June 30, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 16 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
Mistral Large 4 ahead on 5 of 7 counted
Claude Sonnet 5 ahead on 3 of 3 counted
Split 1–1 of 2 counted
Claude Sonnet 5 ahead on 1 of 1 counted
Mistral Large 4 ahead on 3 of 5 counted
No shared comparable results for Abstract reasoning, Computer & tool use, Research & search, Long context, Cybersecurity.
Published scores, with matching modes where available. Gaps show Mistral Large 4 relative to Claude Sonnet 5. Reading the results
| Benchmark | Mistral Large 4 | Claude Sonnet 5 | Gap |
|---|---|---|---|
| Code MigrationMaintenance & Optimization · Hidden-test pass rate (%) | 30.6Thinking High · tools | 44.4≠Max · tools | −13.8 pts#43 vs #24 of 79 |
| DeepSWERepository Engineering · Pass@1 (DeepSWE v1.1) | 61.7Thinking · tools | 54.0≠Deep Thinking · tools | +7.7 pts#45 vs #55 of 73 |
| FrontierCode 1.1Code Generation & Editing · Weighted rubric score (%)Not counted · Different or unrecorded test environments — excluded from the tally | 39.4Thinking High · tools· frontiercode-1-1-main | 56.2≠Extra-High · tools· frontiercode-1-1-extended | —#30 vs #11 of 51 |
| IOI (Vals v2)Algorithmic Coding · Subtask points (%) | 45.3Thinking High · tools | 45≠Max · tools | +0.3 pts#32 vs #33 of 45 |
| Terminal-Bench 4.0Agentic Development · Resolution rate (%) | 28.3Thinking · tools | 12.4≠Max · tools | +15.9 pts#40 vs #59 of 90 |
| Terminal-Bench 4.1Agentic Development · Resolution rate (%) | 27.6Max · tools | 12.1Max · tools | +15.5 pts#26 vs #41 of 48 |
| Vibe Code Bench 1-100Agentic Development · Accuracy (%) | 14.3Thinking High · tools | 13.8≠Max · tools | +0.4 pts#20 vs #21 of 25 |
| Vibe Code Bench v1.1Code Generation & Editing · Pass rate (%) | 78.4Thinking High · tools | 81.3≠Max · tools | −2.9 pts#36 vs #31 of 121 |
| FrontierMath Tier 4 v2Mathematics · Accuracy (verification_code) | 26.8Thinking High · tools | 29.3≠Max · tools | −2.4 pts#40 vs #37 of 73 |
| FrontierMath v2Mathematics · Accuracy (verification_code) | 58.6Thinking High · tools | 65.6≠Max · tools | −7.0 pts#46 vs #38 of 85 |
| ProofBench v1.1 (Lean 4)Mathematics · Accuracy (%) | 10.0Thinking High · tools | 77.0≠Max · tools | −67.0 pts#49 vs #19 of 54 |
| MedCodeClinical Workflows · Accuracy (%) | 40.7Thinking High · no tools | 47.5≠Max · tools | −6.9 pts#58 vs #32 of 97 |
| MedScribeClinical Workflows · Rubric pass rate (%) | 80.4Thinking High · no tools | 76.0≠Max · tools | +4.4 pts#53 vs #71 of 99 |
| MMMU-ProVisual Understanding · Accuracy | 76.4Thinking · no tools | 77.3≠Max · no tools | −0.9 pts#55 vs #49 of 136 |
| EMB (Excel Modeling)Finance · Accuracy (%) | 56.2Thinking High · tools | 66.3≠Max · tools | −10.1 pts#49 vs #24 of 78 |
| Finance Agent v2Finance · Score | 54.7Thinking High · tools | 53.9≠Max · tools | 1.01×#24 vs #29 of 75 |
| Harvey's Legal Agent BenchmarkLegal · Accuracy (%) | 15.8Thinking High · tools | 5.0≠Max · tools | +10.8 pts#6 vs #28 of 52 |
| Legal Research BenchLegal · Score | 31.7Thinking High · tools | 41.8≠Max · tools | 0.76×#52 vs #30 of 91 |
| Tax Agent BenchFinance · Accuracy (%, rubric × citation quality) | 63.3Thinking High · tools | 62.3≠Max · tools | +1.0 pts#14 vs #16 of 74 |
| LiveBenchCross-capability Suites · Accuracy | 71.8Thinking High · no tools | 76.0≠Extra-High · no tools | −4.2 pts#53 vs #29 of 106 |
| LiveBench 2.0Cross-capability Suites · Overall score (equal-weight mean of 7 suite scores, 0-100) | 47.7Thinking High · tools | 73.0≠Max · task tools | 0.65×#47 vs #23 of 51 |
| Vals Index v2.1Capability Indices · GDP-weighted accuracy (%) | 48.0Thinking High · tools | 51.8≠Max · task tools | −3.7 pts#30 vs #21 of 38 |
Maintenance & Optimization · Hidden-test pass rate (%)
Repository Engineering · Pass@1 (DeepSWE v1.1)
Code Generation & Editing · Weighted rubric score (%)
Not counted · Different or unrecorded test environments — excluded from the tally
Algorithmic Coding · Subtask points (%)
Agentic Development · Resolution rate (%)
Agentic Development · Resolution rate (%)
Agentic Development · Accuracy (%)
Code Generation & Editing · Pass rate (%)
Mathematics · Accuracy (verification_code)
Mathematics · Accuracy (verification_code)
Mathematics · Accuracy (%)
Visual Understanding · Accuracy
Finance · Accuracy (%)
Finance · Score
Legal · Accuracy (%)
Legal · Score
Finance · Accuracy (%, rubric × citation quality)
Cross-capability Suites · Accuracy
Cross-capability Suites · Overall score (equal-weight mean of 7 suite scores, 0-100)
Capability Indices · GDP-weighted accuracy (%)
A dash means no result is available.
| Benchmark | Mistral Large 4 | Claude Sonnet 5 |
|---|---|---|
| CritPt | — | 16.9 |
| GPQA Diamond | — | 90.5 |
| HLE | — | 57.4 |
| HLE-Verified | — | 31.0 |
| MysteryMechanism | 12.2 | — |
| SciCode | — | 54.3 |
| AA Cyber Index | 49.5 | — |
| APEX-SWE | — | 46.4 |
| CursorBench 4.0 | — | 34.1 |
| FrontierCode 2.0 | 8.8 | — |
| GSO | — | 37.3 |
| SWE-Atlas-QnA | 59.4 | — |
| SWE-Bench Pro V2 | — | 93.2 |
| SWE-Bench Pro V2 Hard | — | 88.2 |
| SWE-bench Verified | — | 85.2 |
| Terminal-Bench 2.1 | — | 80.4 |
| WeirdML v2 | — | 68.8 |
| WeirdML v3 | 0.044 | — |
| APEX-Agents | — | 54.5 |
| OSWorld-Verified | — | 81.2 |
| Toolathlon-Verified | — | 71.6 |
| AA Intelligence Index v5.0.1 | 25.3 | — |
| Agent Arena | — | 2 |
| Agent Arena Index | — | 1110 |
| ECI | — | 156 |
| Text Arena (Coding) | — | 1544 |
| Vals Index | — | 59.6 |
| Creative Writing | — | 1798 |
| Context Arena | — | 79.5 |
| GDP.pdf | — | 13.2 |
| GeoBench ACW | — | 74.0 |
| AA-AnalystAgent | — | 46.3 |
| AutomationBench | — | 10.7 |
| Harvey Lab-AA | — | 90.1 |
| Public Benefits Bench v1.1 | — | 66.0 |
| SAGE | — | 48.9 |
| τ³-Banking | — | 37.3 |
| SimpleBench | — | 60.6 |
| Vending-Bench 2 | — | 6378 |
| BrowseComp | — | 84.7 |
| BioMysteryBench | 67.0 | — |
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix · 5% steps
Blended price = effective input rate at 90% cache hits × 80% + output rate × 20%
Mistral Large 4 costs77%lessthan Claude Sonnet 5 per million tokens — 4.4× cheaper.Using standard rates at this token mix.
Mistral Large 4: Standard pricing. Mistral official pricing (docs.mistral.ai/inference/pricing) verified 2026-11-13; price reduced from $1.36 Mistral official pricing (docs.mistral.ai/inference/pricing) verified 2026-11-13; price reduced from $4.18
Claude Sonnet 5: Standard pricing. Anthropic official pricing (platform.claude.com/docs/en/about-claude/pricing) verified 2026-11-13; Promotional introductory pricing extended permanently per Anthropic pricing docs
| Price / 1M tokens | Mistral Large 4 | Claude Sonnet 5 | Ratio |
|---|---|---|---|
| Input | $0.70 | $2.00 | 0.35× |
| Output | $2.10 | $10.00 | 0.21× |
| Cached input | $0.07 | $0.20 | 0.35× |
| Cache write | — | — | — |
| Blended | $0.526 | $2.30 | 0.23× |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
Mistral Large 4 relative to Claude Sonnet 5
Summary: 2 of 6 shared benchmarks have matching recorded settings and sources. The average is a geometric mean of 2 positive-cost ratios, one equal weight per benchmark. This benchmark set can include related tests and indexes; it is not a workload cost estimate.
Mistral Large 4 13.4× cheaper
Mistral Large 4 9.5× cheaper
Settings or sources differ
Settings or sources differ
Settings or sources differ
Settings or sources differ
Mistral Large 4 13.4× cheaper
Mistral Large 4 9.5× cheaper
Settings or sources differ
Settings or sources differ
Settings or sources differ
Settings or sources differ
Settings or sources differ or are not recorded; interpret the comparison with care.
Mistral Large 4
Claude Sonnet 5
Settings or sources differ or are not recorded; interpret the comparison with care.
Mistral Large 4
Claude Sonnet 5
Effort, protocol, aggregation and source match; other run conditions may differ.
Mistral Large 4
Claude Sonnet 5
Effort, protocol, aggregation and source match; other run conditions may differ.
Mistral Large 4
Claude Sonnet 5
Settings or sources differ or are not recorded; interpret the comparison with care.
Mistral Large 4
Claude Sonnet 5
Settings or sources differ or are not recorded; interpret the comparison with care.
Mistral Large 4
Claude Sonnet 5
| Benchmark | Mistral Large 4 | Claude Sonnet 5 |
|---|---|---|
| AA Intelligence Index v5.0.1 | Not reported | |
| Agent Arena | Not reported | $0.8542 / task Median · Net improvement index (percentage points vs. average model): 2 Thinking High (With Tools + Internet) |
| Agent Arena Index | Not reported | $0.9521 / task Median · Agent Arena Index rating (index points): 1110 Thinking High (With Tools + Internet) |
| AutomationBench | Not reported | $0.56 / task Average · Pass Rate: 9.9 Thinking High (With Tools) · zapier-private-heldout 2 more settings$0.41 / task Average · Pass Rate: 5.8 Thinking Medium (With Tools) · zapier-private-heldout |
| BioMysteryBench | Not reported | |
| Code Migration | Not reported | |
| CursorBench 4.0 | Not reported | $3.48 / task Average · Task score: 30.8 Thinking High (With Tools) · cursorbench-4-0-cursor-agent $1.39 / task Average · Task score: 24.1 Thinking Low (With Tools) · cursorbench-4-0-cursor-agent 2 more settings$2.31 / task Average · Task score: 28.0 Thinking Medium (With Tools) · cursorbench-4-0-cursor-agent $4.55 / task Average · Task score: 32.0 Extra-High (With Tools) · cursorbench-4-0-cursor-agent |
| DeepSWE | Not reported | 2 more settings |
| EMB (Excel Modeling) | Not reported | |
| Finance Agent v2 | Not reported | |
| FrontierCode 1.1 | Not reported | $12.6849 / task Average · Weighted rubric score: 54.9 Max (With Tools) · frontiercode-1-1-extended $15.4048 / task Average · Weighted rubric score: 42.4 Max (With Tools) · frontiercode-1-1-main $4.4766 / task Average · Weighted rubric score: 52.7 Thinking High (With Tools) · frontiercode-1-1-extended 7 more settings$5.4879 / task Average · Weighted rubric score: 39.4 Thinking High (With Tools) · frontiercode-1-1-main $1.8305 / task Average · Weighted rubric score: 44.5 Thinking Low (With Tools) · frontiercode-1-1-extended $2.1507 / task Average · Weighted rubric score: 28.7 Thinking Low (With Tools) · frontiercode-1-1-main $2.8424 / task Average · Weighted rubric score: 49.3 Thinking Medium (With Tools) · frontiercode-1-1-extended $3.4334 / task Average · Weighted rubric score: 35.2 Thinking Medium (With Tools) · frontiercode-1-1-main $9.0599 / task Average · Weighted rubric score: 42.7 Extra-High (With Tools) · frontiercode-1-1-main $7.4079 / task Average · Weighted rubric score: 56.2 Extra-High (With Tools) · frontiercode-1-1-extended |
| FrontierCode 2.0 | Not reported | |
| Harvey's Legal Agent Benchmark | Not reported | |
| IOI (Vals v2) | Not reported | |
| Legal Research Bench | Not reported | |
| MedCode | Not reported | |
| MedScribe | Not reported | |
| MysteryMechanism | Not reported | |
| ProofBench v1.1 (Lean 4) | Not reported | |
| Vibe Code Bench v1.1 | Not reported | |
| WeirdML v3 | $28.7748 / task Average · Official score (0-1; higher is better): 0.044 Thinking High (With Tools) | Not reported |
| Attribute | Mistral Large 4 | Claude Sonnet 5 |
|---|---|---|
| Availability | Preview | Available |
| Provider | MistralAI | Anthropic |
| Model type | Multimodal model | Multimodal model |
| First release | October 6, 2026 | June 30, 2026 |
| Context window | 1M tokens | 1M tokens |
| Max output | — | 128K tokens |
| Total parameters | 1.1T | — |
| Active parameters | 49B | — |
| Architecture | Mixture of Experts (MoE) | Undisclosed |
| Documented capabilities | Reasoning, Multilingual, Function calling, Tool use, Structured output, Long context, Coding | Reasoning, Multilingual, Coding, Tool use, Function calling, Computer use, Web search, Long context |
| Knowledge cutoff | — | 2026-01 |
| Input modalities | Text, Image | Text, Image |
| Output modalities | Text | Text |
| Weights | Closed / API only | Closed / API only |
| Commercial use | Proprietary | Proprietary |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used Mistral Large 4 or Claude Sonnet 5? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.