GLM-4.7 Flash is Z.ai's free tier: a 200K tokens context model listed at $0 / $0 per Mtok. Inside Atlas, GLM-4.7 Flash has one obvious job, the `small_model` slot, where titles and summaries can be free rather than merely cheap. GLM-4.7 Flash descends from the GLM-4.7 line that costs $0.60 / $2.20 per Mtok, so the free tier is a real model, not a toy. What it is not is a primary build agent: free tiers come with rate limits and no availability guarantee.
What is the Atlas small_model slot and why put GLM-4.7 Flash in it?
GLM-4.7 Flash belongs in the Atlas `small_model` slot, where titles and summaries can be free rather than merely cheap. Set `"small_model": "zai/glm-4.7-flash"` in atlas.json and every background call Atlas makes drops to $0 / $0 per Mtok.
Atlas makes background calls that are not the main agentic work: naming a session, summarizing a thread, small classification tasks. Those calls go to the `small_model`, and on a paid model they quietly accumulate. GLM-4.7 Flash is listed at $0 input and $0 output, so pointing the `small_model` slot at it takes that line to zero. The configuration is one key in atlas.json: `"small_model": "zai/glm-4.7-flash"`. Keep the `model` key on a paid frontier model so the build agent is not rate-limited mid-task, and GLM-4.7 Flash handles only the background work it is suited for.
Is GLM-4.7 Flash actually free?
GLM-4.7 Flash is listed at $0 input and $0 output, with a 200,000 token context window, which no other free hosted tier matches. A free Z.ai account is enough to reach GLM-4.7 Flash, so the only cost is a `ZHIPU_API_KEY`.
Free hosted models are usually small, short-context, or both. GLM-4.7 Flash is neither: it carries a 200,000 token context window at $0 / $0 per Mtok, which no other free hosted tier matches. GLM-4.7 Flash also descends from the GLM-4.7 line that costs $0.60 / $2.20 per Mtok, so the free tier is a real model, not a toy. Export `ZHIPU_API_KEY`, since a free account is enough for this tier, run `atlas models zai` and look for glm-4.7-flash, and Atlas will expose it in `/models` like any other model.
Can GLM-4.7 Flash be your main Atlas model?
GLM-4.7 Flash should not be your main Atlas model. Free tiers come with rate limits and no availability guarantee, so GLM-4.7 Flash is unsuitable as the primary build-agent model, even though its 200,000 token context window would technically hold a large working set.
The risk with GLM-4.7 Flash as a primary model is not capability, it is reliability. A build agent that gets rate-limited mid-task leaves you with a partial change and a stalled session. Atlas computes a unified diff for every file edit and surfaces it for approval before writing, so a stall is recoverable, but it is still a stall. Keep `model` on a paid frontier model so the build agent is not rate-limited mid-task, and keep GLM-4.7 Flash on `small_model` where an occasional retry costs nothing. Atlas lets you switch the active model and provider on the fly, so trying GLM-4.7 Flash on a throwaway task is easy.
What are the data policy limits of GLM-4.7 Flash?
GLM-4.7 Flash is China-hosted, which rules it out under many corporate data policies. That constraint applies even to the `small_model` slot, because Atlas's titles and summaries are derived from your code and conversation, not from generic text.
A free model still sends data somewhere, and for GLM-4.7 Flash that somewhere is China-hosted infrastructure. Many corporate policies rule that out regardless of price, and the `small_model` slot is not an exception, since the summaries and titles Atlas generates are derived from your actual session content. Teams with that constraint have alternatives inside Atlas: Atlas can build its code index with local Ollama embeddings, keeping code off third-party servers, and a paid model in a permitted jurisdiction can take the `small_model` slot instead. If your policy permits it, GLM-4.7 Flash at $0 / $0 per Mtok is the strongest free option available.
When should you pick a different model than GLM-4.7 Flash?
Pick a different model than GLM-4.7 Flash for any work that must complete reliably. GLM-4.7 Flash's free tier comes with rate limits and no availability guarantee, and its China-hosted infrastructure rules it out under many corporate data policies.
The decision rule for GLM-4.7 Flash is simple. Background work that can be retried, and where a delay is harmless: GLM-4.7 Flash at $0 / $0 per Mtok with a 200,000 token context window is the best free option in 2026. Foreground work where a rate limit would strand a half-applied change, or any work under a data policy that excludes China-hosted inference: pick something else. The GLM-4.7 line it descends from costs $0.60 / $2.20 per Mtok, which is a cheap and more reliable step up if you want the same family without the free-tier caveats.
Setup
- 01Export `ZHIPU_API_KEY` (a free account is enough for this tier).
- 02Run `atlas models zai` and look for glm-4.7-flash in the resolved model list.
- 03Set `"small_model": "zai/glm-4.7-flash"` in atlas.json to make background calls free.
- 04Keep `model` on a paid frontier model so the build agent is not rate-limited mid-task.
- 05Confirm GLM-4.7 Flash appears in `/models` before relying on it for background calls.
Frequently asked questions
- Is GLM-4.7 Flash free to use?
- Yes. GLM-4.7 Flash is listed at $0 input and $0 output, with a 200,000 token context window. A free Z.ai account is enough to get a `ZHIPU_API_KEY` for this tier.
- What is the best free model for coding in 2026?
- GLM-4.7 Flash is the most capable genuinely free hosted model available, at $0 / $0 per Mtok on a 200K tokens context window. No other free hosted tier matches that window.
- How do I set GLM-4.7 Flash as Atlas's small_model?
- Set `"small_model": "zai/glm-4.7-flash"` in atlas.json to make background calls free. Keep `model` on a paid frontier model so the build agent is not rate-limited mid-task.
- Can I use GLM-4.7 Flash as my main Atlas model?
- No. Free tiers come with rate limits and no availability guarantee, so GLM-4.7 Flash is unsuitable as the primary build-agent model.
- Where is GLM-4.7 Flash hosted?
- GLM-4.7 Flash is China-hosted, which rules it out under many corporate data policies, including for background calls in the Atlas `small_model` slot.
- What is GLM-4.7 Flash's context window?
- GLM-4.7 Flash has a 200K tokens context window, which is large for a free tier and enough for the summarization and titling work Atlas sends to `small_model`.
- How do I confirm Atlas can see GLM-4.7 Flash?
- Run `atlas models zai` and look for glm-4.7-flash in the resolved list. If it appears there, Atlas will also expose it in the `/models` dialog.
Try Atlas in your terminal
The terminal-native AI coding agent. Free core, single binary.
Install AtlasRelated guides
Atlas for Gin in 2026
Atlas is a terminal-native AI coding agent for Gin in 2026. It reads router groups and binding tags, then runs go test ./... -race behind a permission prompt.
Atlas vs Augment Code: Choosing a Terminal AI Coding Agent in 2026
In 2026, Atlas offers a free, terminal-native AI agent with local embeddings and robust safety. Augment Code's Cosmos platform provides shared memory for teams at $100/month.
Atlas for Assembly: Registers, Calling Conventions, and nasm in 2026
Atlas is a terminal-native AI coding agent for Assembly in 2026. It reads .asm and .S sources, tracks System V and AAPCS64 calling conventions, and assembles with nasm behind a prompt.
Atlas vs Blackbox AI: Choosing Your AI Coding Agent in 2026
Comparing Atlas, the terminal-native AI coding agent, with Blackbox AI, a VS Code agent with 4.7 million installs, for developers in 2026. Evaluate features, pricing, and workflow.
Onboard to an Unfamiliar Codebase with Atlas in 2026
How to onboard to an unfamiliar codebase with Atlas in 2026: use codebase_search, glob, read, lsp, task, and todowrite to build a mental model fast.
Atlas for COBOL: Copybooks, PIC Clauses, and GnuCOBOL in 2026
Atlas is a terminal-native AI coding agent for COBOL in 2026. It reads your divisions and copybooks, explains a paragraph before touching it, and compiles with cobc under GnuCOBOL.
Atlas for Godot: Terminal-Native AI Coding for GDScript and the Node Tree in 2026
Atlas is a terminal-native AI coding agent for Godot in 2026, working across GDScript, the node tree, and signals, where scene structure is half the program.
Plan a Multi-File Change Before Editing with Atlas in 2026
How to plan a multi-file change with Atlas in 2026: the plan agent denies all edit tools, you research with codebase_search and lsp, then plan_exit hands off.