Open-weight Mistral vs GLM
Mistral Large 4 vs GLM-5.3
Mistral Large 4 (Mistral AI) and GLM-5.3 (Z.ai) compared on the facts each vendor publishes. On cost, pick Mistral Large 4: it lists $1.36 in and $4.18 out per 1M tokens, against $1.4 and $4.4 for GLM-5.3. For very long inputs, GLM-5.3 lists the larger context window: 1.05M tokens against 1M. We have not run either model: every value below is Reported, with a link to its source.
Reported Updated · Markdown · All model comparisons
Mistral AI · public preview · open weights · released Oct 6, 2026
- $ in / 1M tokens
- $1.36
- $ out / 1M tokens
- $4.18
- Context
- 1M
- Max output
- —
Z.ai · generally available · open weights · released Aug 14, 2026
- $ in / 1M tokens
- $1.4
- $ out / 1M tokens
- $4.4
- Context
- 1.05M
- Max output
- —
How to choose
- On cost, pick Mistral Large 4: it lists $1.36 in and $4.18 out per 1M tokens, against $1.4 and $4.4 for GLM-5.3.
- For very long inputs, GLM-5.3 lists the larger context window: 1.05M tokens against 1M.
- Both have open weights, under different terms: Mistral Large 4 (Mistral Ecosystem Open-Weight License (MIT derivative with limits on model-as-a-service)) and GLM-5.3 (GLM-5.3 License (MIT-style plus a MaaS condition)). Read each licence before commercial use.
- GLM-5.3 is generally available; Mistral Large 4 is public preview.
- Mistral Large 4 is newer, by 53 days.
Specs side by side
Each value is read from the vendor page linked beside it. “Not published” means we found no vendor figure, not that the feature is missing.
| Spec | Mistral Large 4 | GLM-5.3 |
|---|---|---|
| Price (as listed) | $1.36 in / $4.18 out per 1M tokenssource | $1.4 in / $4.4 out per 1M tokenssource |
| Context window | 1,000,000 tokens (1M)source | 1,048,576 tokens (1.05M)source |
| Output | Not published | textsource |
| Licence | Mistral Ecosystem Open-Weight License (MIT derivative with limits on model-as-a-service) (open weights)source | GLM-5.3 License (MIT-style plus a MaaS condition) (open weights)source |
| Released | Oct 6, 2026source | Aug 14, 2026source |
| Status | public preview | generally available |
| API id | mistral-large-4source | glm-5.3source |
Benchmark snapshots
From the boards named below, not our tests. Boards use different tasks and settings, so compare within a row, never across rows.
| Board | Mistral Large 4 | GLM-5.3 | Type, date |
|---|---|---|---|
| Terminal-Bench 4.0 official leaderboard | not listed | 41.8% | Independent, Oct 26, 2026 |
| DeepSeek-V4.1-Flash-1017 Terminal-Bench 4.0, as reported by DeepSeek | not listed | 37.9% | Vendor-reported, Oct 26, 2026 |
More context: best LLM for coding.
FAQ
Which is cheaper, Mistral Large 4 or GLM-5.3?
Mistral Large 4: $1.36 in / $4.18 out per 1M tokens. GLM-5.3: $1.4 in / $4.4 out per 1M tokens. Both prices are the vendors' own listed prices on the date each model was last verified.
Which has the bigger context window?
GLM-5.3, with 1,048,576 tokens against 1,000,000.
Which is better for coding?
We have not tested either. Sourced snapshots: Terminal-Bench 4.0 official leaderboard (independent, Oct 26, 2026): GLM-5.3 (Claude Code, max) 41.8%; DeepSeek-V4.1-Flash-1017 Terminal-Bench 4.0, as reported by DeepSeek (vendor-reported, Oct 26, 2026): GLM-5.3 37.9%. Scores from different boards are not comparable with each other.
Are Mistral Large 4 and GLM-5.3 open source?
Mistral Large 4: Mistral Ecosystem Open-Weight License (MIT derivative with limits on model-as-a-service), weights downloadable. GLM-5.3: GLM-5.3 License (MIT-style plus a MaaS condition), weights downloadable.
Did you test Mistral Large 4 and GLM-5.3?
No. This page is Reported: every fact is read from the vendor page linked beside it, and benchmark numbers come from the named boards.
Related
Latest news briefs on these models
- Mistral releases the weights of Mistral Large 4 (Le Chonk), a 1T-parameter multimodal MoE model (Reported; first seen 2026-10-29 14:15 UTC)
The Weekly Verdict · free
5 new-model use cases. Every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.