GLM-4.7 vs Codestral 25.08 vs Qwen3 235B-A22B Thinking 2507
Codestral 25.08 comes out ahead, 47 to 42 and 41 on our weighted score, and it is the cheaper option too.
Z.ai (Zhipu)GLM-4.7
41/100- ECI143.7
- Price$0.60 / $2.20
- Context205K
- Our pick
Mistral AICodestral 25.08
47/100- ECI—
- Price$0.30 / $0.90
- Context128K
Alibaba (Qwen)Qwen3 235B-A22B Thinking 2507
42/100- ECI143.8
- Price$0.30 / $2.90
- Context262K
Codestral 25.08 is our pick
Codestral 25.08 is the better all-round choice, scoring 47/100 against Qwen3 235B-A22B Thinking 2507 (42) and GLM-4.7 (41). It leads on price. Qwen3 235B-A22B Thinking 2507 wins on 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 priceCodestral 25.08Codestral 25.08 $0.45 · Qwen3 235B-A22B Thinking 2507 $0.95 · GLM-4.7 $1.00 per 1M tokens (3:1 blend)
- Longest contextQwen3 235B-A22B Thinking 2507Qwen3 235B-A22B Thinking 2507 262,144 · GLM-4.7 204,800 · Codestral 25.08 128,000 tokens
- Widest inputsSame inputsGLM-4.7: Text · Codestral 25.08: Text · Qwen3 235B-A22B Thinking 2507: Text
- Self-hostingGLM-4.7 and Qwen3 235B-A22B Thinking 2507Publishes downloadable weights (Apache 2.0)
| Measure | Weight | GLM-4.7 | Codestral 25.08 | Qwen3 235B-A22B Thinking 2507 |
|---|---|---|---|---|
| Price | 50% | 48 | 64 | 49 |
| Inputs & features | 30% | 35 | 35 | 35 |
| Context window | 20% | 32 | 24 | 37 |
| Overall | 100% | 41/100 | 47/100 | 42/100 |
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 | GLM-4.7Z.ai (Zhipu) | Codestral 25.08Mistral AI | Qwen3 235B-A22B Thinking 2507Alibaba (Qwen) |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 143.7 | — | 143.8 (best) |
| ECI rank | #94 of 165 | — | #93 of 165 (best) |
| GPQA DiamondGraduate-level science questions | 83.3% (best) | — | 80.1% |
| OTIS Mock AIME 2024–2025Competition mathematics | 83.3% | — | 86.7% (best) |
| SimpleQA VerifiedShort factual questions | 32.2% | — | 40.4% (best) |
| Price per million tokens | |||
| Input | $0.60 | $0.30 (best) | $0.30 (best) |
| Output | $2.20 | $0.90 (best) | $2.90 |
| Cached input | $0.11 | $0.03 (best) | — |
| Blended (3:1) | $1.00 | $0.45 (best) | $0.95 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Official Z.AI API | Official Mistral API | Median of 11 providers |
| Limits | |||
| Context window | 204,800 tokens | 128,000 tokens | 262,144 tokens (best) |
| Max output | 131,072 tokens (best) | 8,192 tokens | 81,920 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 | Yes | No | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | No |
| Availability | |||
| Weights | Open | Proprietary | OpenApache 2.0 |
| API model ID | glm-4.7 | codestral-2508 | — |
| API providers | 31 (best) | 11 | 12 |
| Released | Dec 22, 2025 | Jul 30, 2025 | Jul 25, 2025 |
| Knowledge cutoff | Apr 2025 | — | — |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- GLM-4.7$10.40
- Codestral 25.08$4.80
- Qwen3 235B-A22B Thinking 2507$8.80
Which should you choose?
Which is better: GLM-4.7, Codestral 25.08 or Qwen3 235B-A22B Thinking 2507?
Codestral 25.08 is the better all-round choice, scoring 47/100 against Qwen3 235B-A22B Thinking 2507 (42) and GLM-4.7 (41). It leads on price. Qwen3 235B-A22B Thinking 2507 wins on 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, GLM-4.7, Codestral 25.08 or Qwen3 235B-A22B Thinking 2507?
Codestral 25.08 is cheaper at $0.30 input / $0.90 output per million tokens (official Mistral API price). Qwen3 235B-A22B Thinking 2507 costs $0.30 input / $2.90 output per million tokens (median across 11 API providers); GLM-4.7 costs $0.60 input / $2.20 output per million tokens (official Z.AI API price). At a typical mix of three input tokens to one output token, that is $0.45 per million tokens for Codestral 25.08 versus $0.95 for Qwen3 235B-A22B Thinking 2507 (2.1× as much) and $1.00 for GLM-4.7 (2.2× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. GLM-4.7 has an ECI of 143.7, Codestral 25.08 has not been scored yet and Qwen3 235B-A22B Thinking 2507 has an ECI of 143.8.
Which is better for coding?
There are no published SWE-bench Verified results for GLM-4.7, Codestral 25.08 and Qwen3 235B-A22B Thinking 2507 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?
Qwen3 235B-A22B Thinking 2507 has the largest context window at 262,144 tokens, against 204,800 for GLM-4.7 and 128,000 for Codestral 25.08. Maximum output per response: GLM-4.7 up to 131,072, Codestral 25.08 up to 8,192, Qwen3 235B-A22B Thinking 2507 up to 81,920 tokens.
Which can read images, PDFs, audio or video?
GLM-4.7 accepts text; Codestral 25.08 accepts text; Qwen3 235B-A22B Thinking 2507 accepts text. They handle the same number of input types.
Are any of these open source?
GLM-4.7 and Qwen3 235B-A22B Thinking 2507 publishes its weights (Apache 2.0) and can be self-hosted; Codestral 25.08 is proprietary.
Which is newer?
GLM-4.7 is the newest, released Dec 22, 2025. Codestral 25.08 came out Jul 30, 2025; Qwen3 235B-A22B Thinking 2507 came out Jul 25, 2025. Knowledge cutoff: GLM-4.7 Apr 2025.
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 Nov 2, 2026.
