Model comparison
Codestral vs Sonar
Sonar is the stronger model overall, scoring 37.8 to 30.1 on the Noometry Index. Codestral costs 2.2× less per token, which makes it the better buy when Sonar's lead doesn't matter for your workload.
Last verified . 0 shared benchmarks.
Mistral AI
PerplexitySummary
- The widest gap is in coding, where Sonar leads 35.1 to 26.9.
- Codestral is cheaper at $0.30 / $0.90 per million input/output tokens, against $1 / $1 for Sonar.
- Codestral accepts more context: 256K tokens versus 128K.
- Codestral has downloadable open weights; the other is API-only.
Side by side
| Codestral | Sonar | |
|---|---|---|
| Provider | Mistral AI | Perplexity |
| Noometry Index | 30.1 | 37.8 |
| Released | 2024-05-29 | 2024-01-01 |
| Weights | Open | Proprietary |
| Context window | 256K | 128K |
| Max output | 4K | 4K |
| Input $ / M tokens | $0.30 | $1 |
| Output $ / M tokens | $0.90 | $1 |
| Results tracked | 7 | 7 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Sonar leads
Codestral: 26.9 (#316), Sonar: 35.1 (#217)
| Benchmark | Codestral | Sonar |
|---|---|---|
| Aider Polyglot | 11.1% | — |
| BigCodeBench Instruct | 41.8% | — |
| LiveBench Coding | — | 35.1% |
| BigCodeBench Complete | 52.5% | — |
| ALE-Bench | 137.78 | — |
| HumanEval+ | 73.8% | — |
| MBPP+ | 61.9% | — |
Reasoning Sonar leads
Codestral: 19.1 (#244), Sonar: 20.3 (#221)
| Benchmark | Codestral | Sonar |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | — |
| LiveBench Reasoning | — | 46.3% |
| LiveBench Data Analysis | — | 37.9% |
| LiveBench | — | 46.9% |
Math Not comparable
Codestral: —, Sonar: 31.3 (#197)
| Benchmark | Codestral | Sonar |
|---|---|---|
| LiveBench Math | — | 41.6% |
Instruction Following Not comparable
Codestral: —, Sonar: 71.5 (#146)
| Benchmark | Codestral | Sonar |
|---|---|---|
| LiveBench Instruction Following | — | 76.2% |
Writing & Preference Not comparable
Codestral: —, Sonar: 52.6 (#166)
| Benchmark | Codestral | Sonar |
|---|---|---|
| LiveBench Language | — | 44.1% |
Frequently asked questions
Is Codestral better than Sonar?
Sonar is the stronger model overall, scoring 37.8 to 30.1 on the Noometry Index. Codestral costs 2.2× less per token, which makes it the better buy when Sonar's lead doesn't matter for your workload.
Which is cheaper, Codestral or Sonar?
Codestral is cheaper. It lists at $0.30 per million input tokens and $0.90 per million output tokens; Sonar lists at $1 and $1.
Is Codestral or Sonar better for coding?
Sonar scores higher on coding benchmarks: 35.1 versus 26.9 in the Noometry coding category.
Which has the bigger context window?
Codestral does, with 256K tokens against 128K.
How many benchmarks do Codestral and Sonar share?
0 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and Sonar has 7.