Model comparison
Codestral vs Gemini 2.5 Flash
Gemini 2.5 Flash is the stronger model overall, scoring 39.3 to 30.6 on the Noometry Index. Codestral costs 1.9× less per token, which makes it the better buy when Gemini 2.5 Flash's lead doesn't matter for your workload.
Last verified . 3 shared benchmarks.
Mistral AI
GoogleSummary
- They share 3 benchmarks with published results for both. Codestral scores higher in 1 category and Gemini 2.5 Flash in 1 category; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where Gemini 2.5 Flash leads 35.8 to 27.3.
- The biggest single-benchmark swing is Aider Polyglot: 11.1% for Codestral and 55.1% for Gemini 2.5 Flash.
- Codestral is cheaper at $0.30 / $0.90 per million input/output tokens, against $0.30 / $2.50 for Gemini 2.5 Flash.
- Gemini 2.5 Flash accepts more context: 1.05M tokens versus 256K.
Side by side
| Codestral | Gemini 2.5 Flash | |
|---|---|---|
| Provider | Mistral AI | |
| Noometry Index | 30.6 | 39.3 |
| Released | 2024-05-29 | 2025-04-17 |
| Weights | Proprietary | Proprietary |
| Context window | 256K | 1.05M |
| Max output | 8K | 66K |
| Input $ / M tokens | $0.30 | $0.30 |
| Output $ / M tokens | $0.90 | $2.50 |
| Results tracked | 7 | 54 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 2.5 Flash leads
Codestral: 27.3 (#321), Gemini 2.5 Flash: 35.8 (#220)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| Aider Polyglot | 11.1% | 55.1% |
| ALE-Bench | 137.78 | 661.88 |
| SWE-bench Verified (bash only) | — | 28.7% |
| WeirdML | — | 41.9% |
| BigCodeBench Instruct | 41.8% | — |
| LMArena Coding | — | 1424 |
| BigCodeBench Complete | 52.5% | — |
| HumanEval+ | 73.8% | — |
| MBPP+ | 61.9% | — |
Agentic & Tool Use Not comparable
Codestral: —, Gemini 2.5 Flash: 30.8 (#74)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| Terminal-Bench | — | 17.1% |
| Berkeley Function Calling Leaderboard | — | 56.2% |
| TheAgentCompany | — | 41.1% |
| BALROG | — | 33.5% |
| Vending-Bench 2 | — | 548.84 |
Reasoning Codestral leads
Codestral: 19.8 (#251), Gemini 2.5 Flash: 18.1 (#286)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | 56.8% |
| ARC-AGI-2 | — | 2.5% |
| SimpleBench | — | 41.2% |
| ARC-AGI-1 | — | 33.3% |
| CritPt | — | 1.1% |
| EnigmaEval | — | 2.7% |
| LMArena Hard Prompts | — | 1422 |
| DTBench | — | 76.5% |
| LMCA | — | 27.5% |
| Epoch Capabilities Index | — | 143.03 |
| ForecastBench | — | 60.6 |
Math Not comparable
Codestral: —, Gemini 2.5 Flash: 39.9 (#98)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 73.1% |
| Omni-MATH | — | 38.5% |
| LMArena Math | — | 1415 |
| FrontierMath (Feb 2025 set) | — | 4.8% |
| FrontierMath Tier 4 (v1) | — | 4.2% |
Knowledge Not comparable
Codestral: —, Gemini 2.5 Flash: 36.4 (#168)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| Humanity's Last Exam | — | 12.1% |
| MMLU-Pro | — | 63.9% |
| Confabulations | — | 16.8% |
| Vectara Hallucination Rate | — | 7.8% |
| GPQA (HELM) | — | 39% |
| LMArena Expert | — | 1426 |
Multimodal Not comparable
Codestral: —, Gemini 2.5 Flash: 41.8 (#32)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| LMArena Vision | — | 1253 |
| GeoBench | — | 76% |
| VPCT | — | 46.2% |
| SpatialViz-Bench | — | 36.9% |
Multilingual Not comparable
Codestral: —, Gemini 2.5 Flash: 52.3 (#88)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| LMArena Non-English | — | 1409 |
| LMArena Chinese | — | 1450 |
| LMArena French | — | 1433 |
| LMArena German | — | 1418 |
| LMArena Japanese | — | 1405 |
| LMArena Korean | — | 1385 |
| LMArena Russian | — | 1415 |
| LMArena Spanish | — | 1421 |
Instruction Following Not comparable
Codestral: —, Gemini 2.5 Flash: 75.7 (#54)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| IFEval | — | 89.8% |
| LMArena Instruction Following | — | 1405 |
Long Context Not comparable
Codestral: —, Gemini 2.5 Flash: 47.5 (#17)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| Fiction.LiveBench | — | 77.8% |
| LMArena Longer Query | — | 1419 |
Writing & Preference Not comparable
Codestral: —, Gemini 2.5 Flash: 53.8 (#157)
| Benchmark | Codestral | Gemini 2.5 Flash |
|---|---|---|
| LMArena Text | — | 1417 |
| LMArena Creative Writing | — | 1400 |
| Short-Story Creative Writing | — | 76.5% |
| EQ-Bench Creative Writing | — | 1137 |
| WildBench | — | 81.7% |
| LMArena Multi-Turn | — | 1408 |
Frequently asked questions
Is Codestral better than Gemini 2.5 Flash?
Gemini 2.5 Flash is the stronger model overall, scoring 39.3 to 30.6 on the Noometry Index. Codestral costs 1.9× less per token, which makes it the better buy when Gemini 2.5 Flash's lead doesn't matter for your workload.
Which is cheaper, Codestral or Gemini 2.5 Flash?
Codestral is cheaper. It lists at $0.30 per million input tokens and $0.90 per million output tokens; Gemini 2.5 Flash lists at $0.30 and $2.50.
Is Codestral or Gemini 2.5 Flash better for coding?
Gemini 2.5 Flash scores higher on coding benchmarks: 35.8 versus 27.3 in the Noometry coding category.
Which has the bigger context window?
Gemini 2.5 Flash does, with 1.05M tokens against 256K.
How many benchmarks do Codestral and Gemini 2.5 Flash share?
3 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and Gemini 2.5 Flash has 54.