Qwen Flash vs Codestral 25.08 vs ERNIE 4.5 21B A3B
Qwen Flash comes out ahead, 69 to 57 and 47 on our weighted score, though ERNIE 4.5 21B A3B is 11% cheaper per token.
- Our pick
Alibaba (Qwen)Qwen Flash
69/100- ECI—
- Price$0.05 / $0.40
- Context1M
Mistral AICodestral 25.08
47/100- ECI—
- Price$0.30 / $0.90
- Context128K
BaiduERNIE 4.5 21B A3B
57/100- ECI—
- Price$0.07 / $0.28
- Context131K
Qwen Flash is our pick
Qwen Flash is the better all-round choice, scoring 69/100 against ERNIE 4.5 21B A3B (57) and Codestral 25.08 (47). It leads on inputs & features and context window. ERNIE 4.5 21B A3B wins on price. 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 priceERNIE 4.5 21B A3BERNIE 4.5 21B A3B $0.122 · Qwen Flash $0.138 · Codestral 25.08 $0.45 per 1M tokens (3:1 blend)
- Longest contextQwen FlashQwen Flash 1,000,000 · ERNIE 4.5 21B A3B 131,072 · Codestral 25.08 128,000 tokens
- Widest inputsSame inputsQwen Flash: Text · Codestral 25.08: Text · ERNIE 4.5 21B A3B: Text
- Self-hostingERNIE 4.5 21B A3BPublishes downloadable weights (Apache-2.0)
| Measure | Weight | Qwen Flash | Codestral 25.08 | ERNIE 4.5 21B A3B |
|---|---|---|---|---|
| Price | 50% | 87 | 64 | 89 |
| Inputs & features | 30% | 45 | 35 | 25 |
| Context window | 20% | 60 | 24 | 25 |
| Overall | 100% | 69/100 | 47/100 | 57/100 |
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 | Qwen FlashAlibaba (Qwen) | Codestral 25.08Mistral AI | ERNIE 4.5 21B A3BBaidu |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | — | — |
| ECI rank | — | — | — |
| Price per million tokens | |||
| Input | $0.05 (best) | $0.30 | $0.07 |
| Output | $0.40 | $0.90 | $0.28 (best) |
| Cached input | $0.01 (best) | $0.03 | — |
| Blended (3:1) | $0.138 | $0.45 | $0.122 (best) |
| Long-context rate | Over 256K: $0.25 / $2.00 | Same rate | Same rate |
| Price source | Official Alibaba API | Official Mistral API | Median of 1 providers |
| Limits | |||
| Context window | 1,000,000 tokens (best) | 128,000 tokens | 131,072 tokens |
| Max output | 32,768 tokens (best) | 8,192 tokens | 8,000 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 | No |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | Yes | No |
| Availability | |||
| Weights | Proprietary | Proprietary | OpenApache-2.0 |
| API model ID | qwen-flash | codestral-2508 | — |
| API providers | 6 | 11 (best) | 1 |
| Released | Jul 28, 2025 | Jul 30, 2025 | Jun 30, 2025 |
| Knowledge cutoff | 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.
- Qwen Flash$1.30
- Codestral 25.08$4.80
- ERNIE 4.5 21B A3B$1.26
Which should you choose?
Which is better: Qwen Flash, Codestral 25.08 or ERNIE 4.5 21B A3B?
Qwen Flash is the better all-round choice, scoring 69/100 against ERNIE 4.5 21B A3B (57) and Codestral 25.08 (47). It leads on inputs & features and context window. ERNIE 4.5 21B A3B wins on price. 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, Qwen Flash, Codestral 25.08 or ERNIE 4.5 21B A3B?
ERNIE 4.5 21B A3B is cheaper at $0.07 input / $0.28 output per million tokens (median across 1 API provider). Qwen Flash costs $0.05 input / $0.40 output per million tokens (official Alibaba API price); Codestral 25.08 costs $0.30 input / $0.90 output per million tokens (official Mistral API price). At a typical mix of three input tokens to one output token, that is $0.122 per million tokens for ERNIE 4.5 21B A3B versus $0.138 for Qwen Flash (1.1× as much) and $0.45 for Codestral 25.08 (3.7× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Qwen Flash has not been scored yet, Codestral 25.08 has not been scored yet and ERNIE 4.5 21B A3B has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen Flash, Codestral 25.08 and ERNIE 4.5 21B A3B 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?
Qwen Flash has the largest context window at 1,000,000 tokens, against 131,072 for ERNIE 4.5 21B A3B and 128,000 for Codestral 25.08. Maximum output per response: Qwen Flash up to 32,768, Codestral 25.08 up to 8,192, ERNIE 4.5 21B A3B up to 8,000 tokens.
Which can read images, PDFs, audio or video?
Qwen Flash accepts text; Codestral 25.08 accepts text; ERNIE 4.5 21B A3B accepts text. They handle the same number of input types.
Are any of these open source?
ERNIE 4.5 21B A3B publishes its weights (Apache-2.0) and can be self-hosted; Qwen Flash and Codestral 25.08 is proprietary.
Which is newer?
Codestral 25.08 is the newest, released Jul 30, 2025. Qwen Flash came out Jul 28, 2025; ERNIE 4.5 21B A3B came out Jun 30, 2025. Knowledge cutoff: Qwen Flash 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 Nov 2, 2026.
