Across 27 comparable benchmarks, Gemini 3.8 Flash has better reported results on 20 and worse on 7. 1 further shared result is listed below, each with the reason it is not counted.
Strong evidence27 of 28 shared results countedHow we compare
Price basis: Current effective rates. The price section below shows the applicable periods.
| Fact | Gemini 3.8 Flash | Mistral Large 4 |
|---|---|---|
| Context window | 1M tokens | 1M tokens |
| Input / 1M | $0.750 | $0.700 |
| Output / 1M | $3.75 | $2.10 |
| Blended (4:1) | $0.432 | $0.526 |
| Status | Available | Preview |
| Released | September 2, 2026 | October 6, 2026 |
Blended is 4:1 input to output with 90% cache hits at listed rates — set your own mix.
Mean of 19 shared benchmarks grouped by capability on a 0–100 scale. How averages are built
Gemini 3.8 Flash ahead on 6 of 10 counted
Gemini 3.8 Flash ahead on 2 of 3 counted
Gemini 3.8 Flash ahead on 3 of 4 counted
Gemini 3.8 Flash ahead on 1 of 1 counted
Gemini 3.8 Flash ahead on 4 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 Gemini 3.8 Flash relative to Mistral Large 4. Reading the results
| Benchmark | Gemini 3.8 Flash | Mistral Large 4 | Gap |
|---|---|---|---|
| AA Cyber IndexRepository Engineering · Composite index (0–100) | 25.1Thinking High · tools | 49.5≠Thinking · tools | 0.51×#17 vs #7 of 22 |
| Code MigrationMaintenance & Optimization · Hidden-test pass rate (%) | 36.5Thinking High · tools | 30.6Thinking High · tools | +6.0 pts#37 vs #43 of 79 |
| DeepSWERepository Engineering · Pass@1 (DeepSWE v1.1) | 73.8Thinking High · tools | 61.7≠Thinking · tools | +12.1 pts#14 vs #45 of 73 |
| FrontierCode 1.1Code Generation & Editing · Weighted rubric score (%)Not counted · Different or unrecorded test environments — excluded from the tally | 56.3Unspecified Reasoning · tools· frontiercode-1-1-extended | 39.4≠Thinking High · tools· frontiercode-1-1-main | —#10 vs #30 of 51 |
| IOI (Vals v2)Algorithmic Coding · Subtask points (%) | 56.9Thinking High · tools | 45.3Thinking High · tools | +11.7 pts#20 vs #32 of 45 |
| SWE-Atlas-QnARepository Engineering · Task Resolve Rate (%) | 47.0Thinking · tools | 59.4Thinking · tools | −12.4 pts#11 vs #4 of 17 |
| Terminal-Bench 4.0Agentic Development · Resolution rate (%) | 19.7Thinking Medium · tools | 28.3≠Thinking · tools | −8.6 pts#49 vs #40 of 90 |
| Terminal-Bench 4.1Agentic Development · Resolution rate (%) | 18.2Thinking High · tools | 27.6≠Max · tools | −9.4 pts#33 vs #26 of 48 |
| Vibe Code Bench 1-100Agentic Development · Accuracy (%) | 18.8Thinking High · tools | 14.3Thinking High · tools | +4.5 pts#13 vs #20 of 25 |
| Vibe Code Bench v1.1Code Generation & Editing · Pass rate (%) | 78.7Thinking High · tools | 78.4Thinking High · tools | +0.3 pts#35 vs #36 of 121 |
| WeirdML v3ML Engineering · Official score (0-1; higher is better) | 0.074Thinking High · tools | 0.044Thinking High · tools | 1.66×#19 vs #25 of 26 |
| FrontierMath Tier 4 v2Mathematics · Accuracy (verification_code) | 22.0Thinking High · tools | 26.8Thinking High · tools | −4.9 pts#49 vs #40 of 73 |
| FrontierMath v2Mathematics · Accuracy (verification_code) | 68.4Thinking High · tools | 58.6Thinking High · tools | +9.8 pts#33 vs #46 of 85 |
| ProofBench v1.1 (Lean 4)Mathematics · Accuracy (%) | 48.0Thinking High · tools | 10.0Thinking High · tools | +38.0 pts#35 vs #49 of 54 |
| MysteryMechanismScientific Reasoning · Accuracy (%) | 36.5Thinking High · tools | 12.2Thinking High · tools | +24.3 pts#11 vs #32 of 32 |
| BioMysteryBenchBiology & Genomics · Accuracy (%) | 62.2Thinking High · tools | 67.0Thinking High · tools | −4.8 pts#24 vs #21 of 28 |
| MedCodeClinical Workflows · Accuracy (%) | 48.1Thinking High · tools | 40.7≠Thinking High · no tools | +7.4 pts#30 vs #58 of 97 |
| MedScribeClinical Workflows · Rubric pass rate (%) | 84.5Thinking High · tools | 80.4≠Thinking High · no tools | +4.1 pts#34 vs #53 of 99 |
| MMMU-ProVisual Understanding · Accuracy | 85.6Thinking High · no tools | 76.4≠Thinking · no tools | +9.2 pts#7 vs #55 of 136 |
| EMB (Excel Modeling)Finance · Accuracy (%) | 72.2Thinking High · tools | 56.2Thinking High · tools | +16.0 pts#10 vs #49 of 78 |
| Finance Agent v2Finance · Score | 61.4Thinking · tools | 54.7≠Thinking High · tools | 1.12×#2 vs #24 of 75 |
| Harvey's Legal Agent BenchmarkLegal · Accuracy (%) | 10.0Thinking High · tools | 15.8Thinking High · tools | −5.8 pts#13 vs #6 of 52 |
| Legal Research BenchLegal · Score | 38.9Thinking High · tools | 31.7Thinking High · tools | 1.23×#39 vs #52 of 91 |
| Tax Agent BenchFinance · Accuracy (%, rubric × citation quality) | 66.8Thinking High · tools | 63.3Thinking High · tools | +3.5 pts#9 vs #14 of 74 |
| AA Intelligence Index v5.0.1Capability Indices · Index (0-100) | 30.8Thinking High · tools | 25.3≠Thinking · tools | 1.22×#29 vs #39 of 62 |
| LiveBenchCross-capability Suites · Accuracy | 75.8Thinking High · no tools | 71.8Thinking High · no tools | +4.0 pts#30 vs #53 of 106 |
| LiveBench 2.0Cross-capability Suites · Overall score (equal-weight mean of 7 suite scores, 0-100) | 67.3Thinking High · tools | 47.7Thinking High · tools | 1.41×#29 vs #47 of 51 |
| Vals Index v2.1Capability Indices · GDP-weighted accuracy (%) | 54.8Thinking High · tools | 48.0Thinking High · tools | +6.8 pts#15 vs #30 of 38 |
Repository Engineering · Composite index (0–100)
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 (%)
Repository Engineering · Task Resolve Rate (%)
Agentic Development · Resolution rate (%)
Agentic Development · Resolution rate (%)
Agentic Development · Accuracy (%)
Code Generation & Editing · Pass rate (%)
ML Engineering · Official score (0-1; higher is better)
Mathematics · Accuracy (verification_code)
Mathematics · Accuracy (verification_code)
Mathematics · Accuracy (%)
Scientific Reasoning · Accuracy (%)
Biology & Genomics · Accuracy (%)
Clinical Workflows · Accuracy (%)
Clinical Workflows · Rubric pass rate (%)
Visual Understanding · Accuracy
Finance · Accuracy (%)
Finance · Score
Legal · Accuracy (%)
Legal · Score
Finance · Accuracy (%, rubric × citation quality)
Capability Indices · Index (0-100)
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 | Gemini 3.8 Flash | Mistral Large 4 |
|---|---|---|
| CritPt | 18.3 | — |
| FrontierMath Erdős | 0.0 | — |
| GPQA Diamond | 95.3 | — |
| HLE | 44.5 | — |
| HLE-Verified | 54.9 | — |
| Riemann-bench | 51.2 | — |
| SciCode | 56.6 | — |
| Terminal-Bench-Science 0.1 | 12.4 | — |
| APEX-SWE | 36.3 | — |
| CursorBench 4.0 | 39.6 | — |
| FrontierCode 2.0 | — | 8.8 |
| FrontierSWE v2 | 19.6 | — |
| SWE-Bench Pro V2 | 94.9 | — |
| SWE-Bench Pro V2 Hard | 58.8 | — |
| Terminal-Bench 2.1 | 89.4 | — |
| APEX-Agents | 64.3 | — |
| OSWorld 2.0 | 59.0 | — |
| Agent Arena | 2.3 | — |
| Agent Arena Index | 1118 | — |
| ECI | 157 | — |
| Vals Index | 62.3 | — |
| Vals Index v2.2 | 56.0 | — |
| Creative Writing | 1744 | — |
| Hypercritica Model Bench | 72.1 | — |
| AA-LCR | 84.0 | — |
| CharXiv RQ | 86.2 | — |
| GDP.pdf | 22.8 | — |
| GeoBench ACW | 86.0 | — |
| LVBench | 87.8 | — |
| AA-AnalystAgent | 46.3 | — |
| AutomationBench | 29.7 | — |
| Public Benefits Bench v1.1 | 65.3 | — |
| SAGE | 35.1 | — |
| τ³-Banking | 45.8 | — |
| ARC-AGI-1 | 98.5 | — |
| ARC-AGI-2 | 89.2 | — |
| ARC-AGI-3 | 35.0 | — |
| GoBench | 1616 | — |
| SimpleBench | 82.4 | — |
| Vending-Bench 2 | 5094 | — |
| BioMysteryBench (Human-difficult) | 56.5 | — |
| BioMysteryBench (Human-solvable) | 88.8 | — |
| LABBench2 | 86.2 | — |
Both models publish peak and off-peak prices. Each period is compared separately using the same token mix; hours follow each provider’s schedule.
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%
Gemini 3.8 Flash costs64%morethan Mistral Large 4 per million tokens.Using current effective rates at this token mix.
Gemini 3.8 Flash: Listed text rates; pricing tier not confirmed.
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
| Price / 1M tokens | Gemini 3.8 Flash | Mistral Large 4 | Ratio |
|---|---|---|---|
| Input | $0.75 | $0.70 | 1.07× |
| Output | $3.75 | $2.10 | 1.79× |
| Cached input | $0.075 | $0.07 | 1.07× |
| Cache write | — | — | — |
| Blended | $0.864 | $0.526 | 1.64× |
Gemini 3.8 Flash costs228%morethan Mistral Large 4 per million tokens — 3.3× the price.Using standard list rates after promotions end at this token mix.
Gemini 3.8 Flash: Standard pricing. Google Gemini API official pricing (ai.google.dev/gemini-api/docs/pricing) verified 2026-10-21; paid tier; standard tier; scheduled price starting 2027-01-01
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
| Price / 1M tokens | Gemini 3.8 Flash | Mistral Large 4 | Ratio |
|---|---|---|---|
| Input | $1.50 | $0.70 | 2.14× |
| Output | $7.50 | $2.10 | 3.57× |
| Cached input | $0.15 | $0.07 | 2.14× |
| Cache write | — | — | — |
| Blended | $1.73 | $0.526 | 3.28× |
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.
Gemini 3.8 Flash relative to Mistral Large 4
Summary: 6 of 7 shared benchmarks have matching recorded settings and sources. The average is a geometric mean of 6 positive-cost ratios, one equal weight per benchmark. This benchmark set can include related tests and indexes; it is not a workload cost estimate.
Gemini 3.8 Flash 7.3× cheaper
Gemini 3.8 Flash 4.4× more expensive
Gemini 3.8 Flash 3.2× more expensive
Gemini 3.8 Flash 2.8× more expensive
Gemini 3.8 Flash 2.4× cheaper
Gemini 3.8 Flash 1.2× more expensive
Settings or sources differ
Gemini 3.8 Flash 7.3× cheaper
Gemini 3.8 Flash 4.4× more expensive
Gemini 3.8 Flash 3.2× more expensive
Gemini 3.8 Flash 2.8× more expensive
Gemini 3.8 Flash 2.4× cheaper
Gemini 3.8 Flash 1.2× more expensive
Settings or sources differ
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 3.8 Flash
Mistral Large 4
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 3.8 Flash
Mistral Large 4
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 3.8 Flash
Mistral Large 4
Settings or sources differ or are not recorded; interpret the comparison with care.
Gemini 3.8 Flash
Mistral Large 4
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 3.8 Flash
Mistral Large 4
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 3.8 Flash
Mistral Large 4
Effort, protocol, aggregation and source match; other run conditions may differ.
Gemini 3.8 Flash
Mistral Large 4
| Benchmark | Gemini 3.8 Flash | Mistral Large 4 |
|---|---|---|
| AA-LCR | Not reported | |
| Agent Arena | $0.7652 / task Median · Net improvement index (percentage points vs. average model): 2.3 Thinking High (With Tools + Internet) | Not reported |
| Agent Arena Index | $0.7486 / task Median · Agent Arena Index rating (index points): 1118 Thinking High (With Tools + Internet) | Not reported |
| ARC-AGI-1 | Not reported | |
| ARC-AGI-2 | $0.2478 / task Average · ARC-AGI-2 Score: 82.9 Thinking Medium (No Tools) | Not reported |
| AutomationBench | $0.62 / task Average · Pass Rate: 29.7 Thinking High (With Tools) · zapier-private-heldout $0.55 / task Average · Pass Rate: 29.7 Thinking Medium (With Tools) · zapier-private-heldout | Not reported |
| BioMysteryBench | Not reported | |
| Code Migration | Not reported | |
| CursorBench 4.0 | $4.70 / task Average · Task score: 39.6 Thinking High (With Tools) · cursorbench-4-0-cursor-agent $4.06 / task Average · Task score: 37.3 Thinking Medium (With Tools) · cursorbench-4-0-cursor-agent | Not reported |
| DeepSWE | Not reported | |
| EMB (Excel Modeling) | Not reported | |
| Finance Agent v2 | Not reported | |
| FrontierCode 2.0 | Not reported | |
| FrontierSWE v2 | 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 | |
| Tax Agent Bench | Not reported | $2.7212 / task Average · Accuracy (%, rubric × citation quality): 63.3 Thinking High (With Tools) |
| Vals Index v2.2 | Not reported | |
| Vibe Code Bench v1.1 | Not reported |
| Attribute | Gemini 3.8 Flash | Mistral Large 4 |
|---|---|---|
| Availability | Available | Preview |
| Provider | Google Deep Mind | MistralAI |
| Model type | Multimodal model | Multimodal model |
| First release | September 2, 2026 | October 6, 2026 |
| Context window | 1M tokens | 1M tokens |
| Max output | 64K tokens | — |
| Total parameters | — | 1.1T |
| Active parameters | — | 49B |
| Architecture | Undisclosed | Mixture of Experts (MoE) |
| Documented capabilities | Reasoning, Coding, Function calling, Tool use, Structured output, Computer use, Web search, Long context | Reasoning, Multilingual, Function calling, Tool use, Structured output, Long context, Coding |
| Input modalities | Text, Image, Audio, Video | 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 Gemini 3.8 Flash or Mistral Large 4? 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.