Codestral vs Mistral Large 2.1 vs Qwen2.5 72B Instruct
Codestral comes out ahead, 48 to 28 and 26 on our weighted score, and it is the cheaper option too.
- Our pick
Mistral AICodestral
48/100- ECI—
- Price$0.30 / $0.90
- Context256K
- Mistral AI
Mistral Large 2.1
26/100- ECI128.5
- Price$2.00 / $6.00
- Context131K
Alibaba (Qwen)Qwen2.5 72B Instruct
28/100- ECI129.0
- Price$1.40 / $5.60
- Context131K
Codestral is our pick
Codestral is the better all-round choice, scoring 48/100 against Qwen2.5 72B Instruct (28) and Mistral Large 2.1 (26). It leads on price and context window. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceCodestralCodestral $0.45 · Qwen2.5 72B Instruct $2.45 · Mistral Large 2.1 $3.00 per 1M tokens (3:1 blend)
- Longest contextCodestralCodestral 256,000 · Mistral Large 2.1 131,072 · Qwen2.5 72B Instruct 131,072 tokens
- Widest inputsSame inputsCodestral: Text · Mistral Large 2.1: Text · Qwen2.5 72B Instruct: Text
- Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
| Measure | Weight | Codestral | Mistral Large 2.1 | Qwen2.5 72B Instruct |
|---|---|---|---|---|
| Price | 50% | 66 | 27 | 31 |
| Inputs & features | 30% | 25 | 25 | 25 |
| Context window | 20% | 36 | 24 | 24 |
| Overall | 100% | 48/100 | 26/100 | 28/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | CodestralMistral AI | Mistral Large 2.1Mistral AI | Qwen2.5 72B InstructAlibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | 128.5 | 129.0 (best) |
| ECI rank | — | #130 of 148 | #128 of 148 (best) |
| GPQA DiamondGraduate-level science questions | — | 51.3% (best) | 49.2% |
| OTIS Mock AIME 2024–2025Competition mathematics | — | 7.8% | 8.1% (best) |
| Price per million tokens | |||
| Input | $0.30 (best) | $2.00 | $1.40 |
| Output | $0.90 (best) | $6.00 | $5.60 |
| Cached input | $0.03 | — | — |
| Blended (3:1) | $0.45 (best) | $3.00 | $2.45 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Mistral API | Official Mistral API | Official Alibaba API |
| Limits | |||
| Context window | 256,000 tokens (best) | 131,072 tokens | 131,072 tokens |
| Max output | 4,096 tokens | 16,384 tokens (best) | 8,192 tokens |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | No | No | No |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | No | No | No |
| Reasoning | No | No | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | No | No |
| Availability | |||
| Weights | Open | Open | Open |
| API model ID | codestral-latest | mistral-large-2411 | qwen2-5-72b-instruct |
| API providers | 3 (best) | 2 | 1 |
| Released | May 29, 2024 | Nov 18, 2024 | Sep 19, 2024 |
| Knowledge cutoff | Oct 2024 | Nov 2024 | Apr 2024 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Codestral$4.80
- Mistral Large 2.1$32.00
- Qwen2.5 72B Instruct$25.20
Which should you choose?
Which is better: Codestral, Mistral Large 2.1 or Qwen2.5 72B Instruct?
Codestral is the better all-round choice, scoring 48/100 against Qwen2.5 72B Instruct (28) and Mistral Large 2.1 (26). It leads on price and context window. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Codestral, Mistral Large 2.1 or Qwen2.5 72B Instruct?
Codestral is cheaper at $0.30 input / $0.90 output per million tokens (official Mistral API price). Qwen2.5 72B Instruct costs $1.40 input / $5.60 output per million tokens (official Alibaba API price); Mistral Large 2.1 costs $2.00 input / $6.00 output per million tokens (official Mistral API price). At a typical mix of three input tokens to one output token, that is $0.45 per million tokens for Codestral versus $2.45 for Qwen2.5 72B Instruct (5.4× as much) and $3.00 for Mistral Large 2.1 (6.7× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Codestral has not been scored yet, Mistral Large 2.1 has an ECI of 128.5 and Qwen2.5 72B Instruct has an ECI of 129.0.
Which is better for coding?
There are no published SWE-bench Verified results for Codestral, Mistral Large 2.1 and Qwen2.5 72B Instruct yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
Codestral has the largest context window at 256,000 tokens, against 131,072 for Mistral Large 2.1 and 131,072 for Qwen2.5 72B Instruct. Maximum output per response: Codestral up to 4,096, Mistral Large 2.1 up to 16,384, Qwen2.5 72B Instruct up to 8,192 tokens.
Which can read images, PDFs, audio or video?
Codestral accepts text; Mistral Large 2.1 accepts text; Qwen2.5 72B Instruct accepts text. They handle the same number of input types.
Are any of these open source?
Yes, all three publish their weights, so you can self-host them.
Which is newer?
Mistral Large 2.1 is the newest, released Nov 18, 2024. Qwen2.5 72B Instruct came out Sep 19, 2024; Codestral came out May 29, 2024. Knowledge cutoff: Codestral Oct 2024, Mistral Large 2.1 Nov 2024, Qwen2.5 72B Instruct Apr 2024.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.
