Model comparison
Codestral vs DeepSeek-R1-Distill-Llama-70B
DeepSeek-R1-Distill-Llama-70B is the stronger model overall, scoring 37.8 to 30.6 on the Noometry Index.
Last verified . 3 shared benchmarks.
Mistral AI
DeepSeekSummary
- They share 3 benchmarks with published results for both. Codestral scores higher in 0 categories and DeepSeek-R1-Distill-Llama-70B in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where DeepSeek-R1-Distill-Llama-70B leads 36.8 to 27.3.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 32.5% for Codestral and 52.3% for DeepSeek-R1-Distill-Llama-70B.
- DeepSeek-R1-Distill-Llama-70B has downloadable open weights; the other is API-only.
Side by side
| Codestral | DeepSeek-R1-Distill-Llama-70B | |
|---|---|---|
| Provider | Mistral AI | DeepSeek |
| Noometry Index | 30.6 | 37.8 |
| Released | 2024-05-29 | 2025-01-20 |
| Weights | Proprietary | Open |
| Context window | 256K | — |
| Max output | 8K | — |
| Input $ / M tokens | $0.30 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 7 | 13 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1-Distill-Llama-70B leads
Codestral: 27.3 (#321), DeepSeek-R1-Distill-Llama-70B: 36.8 (#202)
| Benchmark | Codestral | DeepSeek-R1-Distill-Llama-70B |
|---|---|---|
| BigCodeBench Instruct | 41.8% | 35.3% |
| BigCodeBench Complete | 52.5% | 49.9% |
| Aider Polyglot | 11.1% | — |
| LiveBench Coding | — | 51.6% |
| ALE-Bench | 137.78 | — |
| HumanEval+ | 73.8% | — |
| MBPP+ | 61.9% | — |
Reasoning DeepSeek-R1-Distill-Llama-70B leads
Codestral: 19.8 (#251), DeepSeek-R1-Distill-Llama-70B: 24.9 (#156)
| Benchmark | Codestral | DeepSeek-R1-Distill-Llama-70B |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | 52.3% |
| LiveBench Reasoning | — | 67.6% |
| LiveBench Data Analysis | — | 55.9% |
| LiveBench | — | 54.5% |
Math Not comparable
Codestral: —, DeepSeek-R1-Distill-Llama-70B: 36.0 (#176)
| Benchmark | Codestral | DeepSeek-R1-Distill-Llama-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 51.4% |
| LiveBench Math | — | 58.1% |
| MATH Level 5 | — | 89.9% |
Knowledge Not comparable
Codestral: —, DeepSeek-R1-Distill-Llama-70B: 30.7 (#225)
| Benchmark | Codestral | DeepSeek-R1-Distill-Llama-70B |
|---|---|---|
| GPQA Diamond | — | 55.7% |
Instruction Following Not comparable
Codestral: —, DeepSeek-R1-Distill-Llama-70B: 68.2 (#190)
| Benchmark | Codestral | DeepSeek-R1-Distill-Llama-70B |
|---|---|---|
| LiveBench Instruction Following | — | 69.9% |
Writing & Preference Not comparable
Codestral: —, DeepSeek-R1-Distill-Llama-70B: 49.0 (#194)
| Benchmark | Codestral | DeepSeek-R1-Distill-Llama-70B |
|---|---|---|
| LiveBench Language | — | 23.8% |
Frequently asked questions
Is Codestral better than DeepSeek-R1-Distill-Llama-70B?
DeepSeek-R1-Distill-Llama-70B is the stronger model overall, scoring 37.8 to 30.6 on the Noometry Index.
Is Codestral or DeepSeek-R1-Distill-Llama-70B better for coding?
DeepSeek-R1-Distill-Llama-70B scores higher on coding benchmarks: 36.8 versus 27.3 in the Noometry coding category.
How many benchmarks do Codestral and DeepSeek-R1-Distill-Llama-70B share?
3 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and DeepSeek-R1-Distill-Llama-70B has 13.