Across 14 comparable benchmarks, Claude Haiku 5.5 has better reported results on 14 and worse on 0.
Moderate evidence14 of 14 shared results countedHow we compare
Price basis: Listed rates vs standard rates. The price section below shows the applicable periods.
| Fact | Claude Haiku 5.5 | Mistral Medium 3.5 |
|---|---|---|
| Context window | 1M tokens | 256K tokens |
| Input / 1M | $0.100 | $1.50 |
| Output / 1M | $0.500 | $7.50 |
| Blended (4:1) | $0.115 | $2.70 |
| Status | Available | Available |
| Released | October 7, 2026 | May 1, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates β set your own mix.
Mean of 10 shared benchmarks grouped by capability on a 0β100 scale. How averages are built
Claude Haiku 5.5 ahead on 3 of 3 counted
Claude Haiku 5.5 ahead on 1 of 1 counted
Claude Haiku 5.5 ahead on 2 of 2 counted
Claude Haiku 5.5 ahead on 7 of 7 counted
No shared comparable results for Abstract reasoning, Computer & tool use, Research & search, Long context, Multimodal, Cybersecurity.
Published scores, with matching modes where available. Gaps show Claude Haiku 5.5 relative to Mistral Medium 3.5. Reading the results
| Benchmark | Claude Haiku 5.5 | Mistral Medium 3.5 | Gap |
|---|---|---|---|
| Code MigrationMaintenance & Optimization Β· Hidden-test pass rate (%) | 44.6Max Β· tools | 5.1β Thinking High Β· tools | +39.4 pts#21 vs #72 of 79 |
| FrontierCode 1.1Code Generation & Editing Β· Weighted rubric score (%) | 46.4Max Β· toolsΒ· frontiercode-1-1-main | 8.0β Normal Β· toolsΒ· frontiercode-1-1-main | +38.4 pts#15 vs #51 of 51 |
| Vibe Code Bench v1.1Code Generation & Editing Β· Pass rate (%) | 90.4Max Β· tools | 2.9β Thinking High Β· tools | +87.5 pts#5 vs #115 of 121 |
| ProofBench v1.1 (Lean 4)Mathematics Β· Accuracy (%) | 89.0Max Β· tools | 9.0β Thinking High Β· tools | +80.0 pts#13 vs #50 of 54 |
| MedCodeClinical Workflows Β· Accuracy (%) | 51.5Max Β· tools | 33.8β Thinking High Β· tools | +17.8 pts#15 vs #78 of 97 |
| MedScribeClinical Workflows Β· Rubric pass rate (%) | 89.5Max Β· tools | 67.7β Thinking High Β· tools | +21.8 pts#6 vs #90 of 99 |
| AA-AnalystAgentData Analysis Β· pass^5 (%) | 53.8Max Β· tools + web | 12.5β Normal Β· tools + web | +41.3 pts#7 vs #34 of 39 |
| AA-BriefcaseOffice & Business Β· AA-Briefcase Elo | 1578Max Β· tools + web | 519β Normal Β· tools | 3.04Γ#13 vs #19 of 20 |
| EMB (Excel Modeling)Finance Β· Accuracy (%) | 59.0Max Β· tools | 11.2β Thinking High Β· tools | +47.8 pts#39 vs #76 of 78 |
| Finance Agent v2Finance Β· Score | 54.1Unspecified Reasoning Β· tools | 32.1β Thinking High Β· tools | 1.69Γ#27 vs #67 of 75 |
| Legal Research BenchLegal Β· Score | 43.8Max Β· tools | 9.1β Thinking High Β· tools | 4.79Γ#26 vs #83 of 91 |
| SAGEService Workflows Β· Accuracy (%) | 45.8Max Β· tools | 37.6β Thinking High Β· tools | +8.2 pts#38 vs #62 of 82 |
| Tax Agent BenchFinance Β· Accuracy (%, rubric Γ citation quality) | 31.9Unspecified Reasoning Β· tools | 4.5β Thinking High Β· tools | +27.4 pts#38 vs #73 of 74 |
| AA Intelligence Index v5.0.1Capability Indices Β· Index (0-100) | 36.5Max Β· task tools | 4.6β Thinking High Β· tools | 7.85Γ#20 vs #59 of 62 |
Maintenance & Optimization Β· Hidden-test pass rate (%)
Code Generation & Editing Β· Weighted rubric score (%)
Code Generation & Editing Β· Pass rate (%)
Mathematics Β· Accuracy (%)
Data Analysis Β· pass^5 (%)
Office & Business Β· AA-Briefcase Elo
Finance Β· Accuracy (%)
Finance Β· Score
Legal Β· Score
Service Workflows Β· Accuracy (%)
Finance Β· Accuracy (%, rubric Γ citation quality)
Capability Indices Β· Index (0-100)
A dash means no result is available.
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 Β· 5% steps
Blended price = effective input rate at 90% cache hits Γ 80% + output rate Γ 20%
Claude Haiku 5.5 costs96%lessthan Mistral Medium 3.5 per million tokens β 23.4Γ cheaper.Using listed rates vs standard rates at this token mix.
Claude Haiku 5.5: Standard pricing. context_length <= 100000 Prompt length tier; text and image inputs are token billed. context_length > 100000
Mistral Medium 3.5: Standard pricing. Mistral official pricing verified 2026-11-13; Deprecated 2026-10-31, scheduled retirement 2027-04-30 per docs.mistral.ai/models
| Price / 1M tokens | Claude Haiku 5.5 | Mistral Medium 3.5 | Ratio |
|---|---|---|---|
| Input | $0.10 | $1.50 | 0.07Γ |
| Output | $0.50 | $7.50 | 0.07Γ |
| Cached input | $0.01 | β | β |
| Cache write | $0.125 | β | β |
| Blended | $0.115 | $2.70 | 0.04Γ |
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.
Claude Haiku 5.5 relative to Mistral Medium 3.5
Summary: 1 of 2 shared benchmarks have matching recorded settings and sources. The average is a geometric mean of 1 positive-cost ratios, one equal weight per benchmark. This benchmark set can include related tests and indexes; it is not a workload cost estimate.
Claude Haiku 5.5 6.6Γ cheaper
Settings or sources differ
Claude Haiku 5.5 6.6Γ cheaper
Settings or sources differ
Effort, protocol, aggregation and source match; other run conditions may differ.
Claude Haiku 5.5
Mistral Medium 3.5
Settings or sources differ or are not recorded; interpret the comparison with care.
Claude Haiku 5.5
Mistral Medium 3.5
| Benchmark | Claude Haiku 5.5 | Mistral Medium 3.5 |
|---|---|---|
| AA Decision Index v2.1 | Not reported | |
| Agent Arena | $0.0778 / task Median Β· Net improvement index (percentage points vs. average model): 0.42 Thinking High (With Tools + Internet) | Not reported |
| Agent Arena Index | $0.0944 / task Median Β· Agent Arena Index rating (index points): 1095 Thinking High (With Tools + Internet) | Not reported |
| ARC-AGI-1 | 2 more settings | Not reported |
| ARC-AGI-2 | 1 more settings | Not reported |
| AutomationBench | $0.03 / task Average Β· Pass Rate: 10.5 Thinking High (With Tools) Β· zapier-private-heldout $0.02 / task Average Β· Pass Rate: 6.1 Thinking Low (With Tools) Β· zapier-private-heldout 2 more settings$0.02 / task Average Β· Pass Rate: 8.5 Thinking Medium (With Tools) Β· zapier-private-heldout $0.04 / task Average Β· Pass Rate: 9.9 Extra-High (With Tools) Β· zapier-private-heldout | Not reported |
| Code Migration | Not reported | |
| CursorBench 4.0 | $1.12 / task Average Β· Task score: 48.4 Max (With Tools) Β· cursorbench-4-0-cursor-agent $0.32 / task Average Β· Task score: 42.3 Thinking High (With Tools) Β· cursorbench-4-0-cursor-agent $0.08 / task Average Β· Task score: 30.9 Thinking Low (With Tools) Β· cursorbench-4-0-cursor-agent 2 more settings$0.17 / task Average Β· Task score: 36.9 Thinking Medium (With Tools) Β· cursorbench-4-0-cursor-agent $0.56 / task Average Β· Task score: 44.3 Extra-High (With Tools) Β· cursorbench-4-0-cursor-agent | Not reported |
| EMB (Excel Modeling) | Not reported | |
| Finance Agent v2 | Not reported | |
| FrontierCode 1.1 | $1.3284 / task Average Β· Weighted rubric score: 46.4 Max (With Tools) Β· frontiercode-1-1-main $0.256 / task Average Β· Weighted rubric score: 41.9 Thinking High (With Tools) Β· frontiercode-1-1-main $0.0596 / task Average Β· Weighted rubric score: 34.8 Thinking Low (With Tools) Β· frontiercode-1-1-main 2 more settings$0.13 / task Average Β· Weighted rubric score: 41.6 Thinking Medium (With Tools) Β· frontiercode-1-1-main $0.6297 / task Average Β· Weighted rubric score: 45.8 Extra-High (With Tools) Β· frontiercode-1-1-main | Not reported |
| FrontierCode 2.0 | $0.139 / task Average Β· Weighted rubric score: 18.6 Thinking High (With Tools + Internet) $0.065 / task Average Β· Weighted rubric score: 10.5 Thinking Low (With Tools + Internet) 2 more settings$0.097 / task Average Β· Weighted rubric score: 13.4 Thinking Medium (With Tools + Internet) | Not reported |
| FrontierSWE v2 | Not reported | |
| GDPx | Not reported | |
| IOI (Vals v2) | Not reported | |
| Legal Research Bench | Not reported | |
| LiveBench 2.0 | $0.6189 / task Average Β· Overall score (equal-weight mean of 7 suite scores, 0-100): 75.6 Max (task-specific tools) | Not reported |
| MedCode | Not reported | |
| MedScribe | Not reported | |
| MysteryMechanism | Not reported | |
| ProofBench v1.1 (Lean 4) | Not reported | |
| Public Benefits Bench v1.1 | Not reported | |
| SAGE | Not reported | |
| Terminal-Bench 2.1 | Not reported | |
| Terminal-Bench 4.0 | $3.12 / task Average Β· Resolution rate (%): 36.4 Unspecified Reasoning (With Tools) Β· vals-mini-swe | Not reported |
| Terminal-Bench 4.1 | Not reported | |
| Vals Index v2.2 | Not reported | |
| Vibe Code Bench 1-100 | Not reported | |
| Vibe Code Bench v1.1 | Not reported | |
| WeirdML v3 | $2.3004 / task Average Β· Official score (0-1; higher is better): 0.12 Extra-High (With Tools) | Not reported |
| Attribute | Claude Haiku 5.5 | Mistral Medium 3.5 |
|---|---|---|
| Availability | Available | Available |
| Provider | Anthropic | MistralAI |
| Model type | Reasoning model | Chat model |
| First release | October 7, 2026 | May 1, 2026 |
| Context window | 1M tokens | 256K tokens |
| Max output | 128K tokens | β |
| Architecture | Undisclosed | Undisclosed |
| Documented capabilities | Reasoning | Coding, Tool use, Function calling, Long context |
| Knowledge cutoff | 2026-06 | β |
| Input modalities | Text, Image | Text |
| 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 Claude Haiku 5.5 or Mistral Medium 3.5? 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.