GLM-5.2
MIT-licensed 744B open-weight coding model from Zhipu AI with a 1M-token context, built for long-horizon agentic work you can self-host.
Updated 2026-08-10
Not settled here. This entry does not record a confirmed free tier for GLM-5.2.
Listed pricing: Free chat + MIT open weights; paid API and GLM Coding Plan — check site for current rates.
Overview
Is GLM-5.2 free?
Not settled here. This entry does not record a confirmed free tier for GLM-5.2.
What the entry records: Yes — free chat access plus MIT-licensed weights you can download and run at no license cost. Hosted API and Coding Plan rates are not stated here because they were not verified against z.ai on the date of this update; check the site directly before budgeting.
Listed pricing: Free chat + MIT open weights; paid API and GLM Coding Plan — check site for current rates.
No figure is estimated here for what is not published. Check GLM-5.2's own site for the current terms.
Pricing verified . That is the date the plans were last re-checked against the vendor's own pages, not today's date. Confirm on the official site before you pay.
GLM-5.2 pricing
| Plan | Price | What's included |
|---|---|---|
| Chat (Free) | $0 | Browser access to GLM-5.2 through the Z.ai chat interface, with usage limits. Enough to evaluate the model's coding and reasoning behaviour before committing. |
| Open weights (self-host) | $0 license | Full model weights on Hugging Face under MIT — no fee, no usage reporting, no restrictions on commercial use or fine-tuning. Your cost is entirely hardware and inference operations. |
| API (pay-as-you-go) | Check website for current pricing | Token-based access to Zhipu's hosted endpoint for production workloads. A free token allowance was offered at signup around launch; verify whether it is still running before budgeting around it. |
| GLM Coding Plan | Check website for current pricing | Subscription tier aimed at developers driving GLM from coding agents and IDE integrations, with higher allowances than pay-as-you-go. |
Browser access to GLM-5.2 through the Z.ai chat interface, with usage limits. Enough to evaluate the model's coding and reasoning behaviour before committing.
Full model weights on Hugging Face under MIT — no fee, no usage reporting, no restrictions on commercial use or fine-tuning. Your cost is entirely hardware and inference operations.
Token-based access to Zhipu's hosted endpoint for production workloads. A free token allowance was offered at signup around launch; verify whether it is still running before budgeting around it.
Subscription tier aimed at developers driving GLM from coding agents and IDE integrations, with higher allowances than pay-as-you-go.
2 plans carry no published number: API (pay-as-you-go), recorded as “Check website for current pricing”; GLM Coding Plan, recorded as “Check website for current pricing”. Nothing is estimated in its place. The vendor's own pricing page is the only source for those figures.
Pricing verified . That is the date the plans were last re-checked against the vendor's own pages, not today's date. Confirm on the official site before you pay.
Is GLM-5.2 worth it?
Worth it for Engineering teams building coding agents who need frontier-class capability under a license they fully control — self-hosted, fine-tuned, or air-gapped — rather than rented from a closed API.
There is no free tier to test it on, so the plans above are the entry cost. The recorded trade-offs are listed below, and any one of them can settle the question on its own.
The 8.5/10 AI Score is an editorial read of published capability, price and shipping pace. Nobody here has hands-on hours with GLM-5.2. How we verify.
Worth it if
The strengths recorded against this entry.
- MIT license at 744B parameters — commercial use, fine-tuning and redistribution with no revocable API terms or MAU clause
- 1M-token context handles whole repositories and long agent histories in one pass
- Tuned for long-horizon, multi-turn tool use rather than one-shot completion, which is the workload coding agents actually run
- Self-hosting is a genuine option for teams with data-residency or air-gap requirements, not a theoretical one
- Materially cheaper than closed frontier tiers like GPT-5.6 Sol at $5/$30 per million tokens
Not worth it if
Any one of these blocks your use case.
- 744B parameters means serious multi-GPU hardware to self-host, so most teams end up on the hosted API anyway — back on someone else's terms
- The launch claim of beating GPT-5.5 on long-horizon coding was vendor- and press-reported and still has no independent replication; GPT-5.6 Sol shipped in July and moved the comparison point
- The cheap-frontier field caught up fast — Qwen 3.8 Max at $2/$6 and Kimi K3 at $3/$15 offer comparable context, so GLM-5.2's cost edge is no longer distinctive
- Hosted API and Coding Plan prices are not published in an easily comparable form, and some enterprises face data-residency or procurement blocks on a China-based provider
What sets GLM-5.2 apart
- MIT-licensed weights at 744B parameters, where rivals ship either closed weights or restrictive community licenses
- 1M-token context in a model you can download and run yourself, not just call over an API
- Open weights are shipped, not promised — Qwen 3.8 Max's comparable release was still a commitment as of early August 2026
- Nothing its category rivals lack on raw capability; it competes on licensing terms and deployment control
Key features
744B MIT open weights
The full model is downloadable under an MIT license, so commercial deployment, fine-tuning and redistribution are all permitted with no usage ceiling and no terms that can be revoked mid-contract. At frontier scale that licensing is still uncommon — most rivals ship either closed weights or a restrictive community license.
1M-token context
A one-million-token window lets it hold an entire repository, a long document set, or the accumulated history of a running agent session in a single pass, which is what keeps a multi-hour agent loop coherent instead of drifting once early context falls out.
Long-horizon agentic tuning
Trained for multi-turn work where the model plans, calls tools, reads results and iterates over many steps. This is the axis Zhipu led its launch benchmarks on, and it is a different capability from single-shot code generation, where the open-weight field has been competitive for much longer.
Three access paths
Free chat at chat.z.ai for evaluation, a hosted API for production, and the raw weights for on-premise or air-gapped deployment. The third path is the one that matters for teams with data-residency rules, since nothing has to leave their infrastructure.
How it compares
| Tool | Best for | Pricing | Score |
|---|---|---|---|
| GLM-5.2 | Engineering teams building coding agents who need frontier-class capability under a license they fully control — self-hosted, fine-tuned, or air-gapped — rather than rented from a closed API. | Free chat + MIT open weights; paid API and GLM Coding Plan — check site for current rates | 8.5/10 |
| Claude Code vs Claude Code → | Developers who want an agent that works inside an existing repo and toolchain rather than in a hosted editor, and who already pay for a Claude plan. | Included with Claude Free, Pro, Max, Team and Enterprise plans | 9.3/10 |
| Cursor vs Cursor → | Professional developers who want an IDE-native agent that can also run in the cloud and be steered from a phone, and engineering teams that need admin control over which model handles which request. | Free Hobby + Individual $20/mo + Teams $40/user/mo + Enterprise custom | 9.2/10 |
| v0 by Vercel vs v0 by Vercel → | Frontend developers on React and Next.js who want production-ready UI generated from a text prompt or a screenshot instead of hand-built boilerplate. | Free tier + Plus $30/user/mo + Business $100/user/mo + Enterprise custom | 9/10 |
Compare head-to-head
Related reading
GLM-5.2: Z.ai Open Model Tops Coding Benchmarks
Z.ai's GLM-5.2 open-weights model beats GPT-5.5 on long-horizon coding benchmarks at a fraction of the cost, per VentureBeat.
Muse Glimmer and the 30B Open-Weights Ceiling
Meta released Muse Glimmer's 30B weights under Apache 2.0, landing in the same consumer-GPU size class as every other recent open model release.
OpenAI Astra's Cyber Pause and the Release Rules
OpenAI flagged Astra as critical on cybersecurity and paused non-essential work on it. What the hold covers and what nobody outside OpenAI can check.
Ready to try GLM-5.2?
Head to the official site to start with GLM-5.2 — pricing and plans are listed above.
Visit GLM-5.2