docs: update models for GLM 5.3 support

[tembo]

What changed

Added GLM 5.3 (glm-5p3) and GLM 5.3 Flash (glm-5p3-flash) to the Supported Models table in features/models.mdx, and added both to the Opencode and Pi harness model lists in features/agents.mdx.

Why

The monorepo added these two models to the production model registry, litellm config, pricing table, and the Opencode/Pi harness allow-lists, but the docs' Supported Models table and per-harness model lists still only listed GLM 5.2.

Evidence

• monorepo PR #10871 "feat(models): add Fireworks GLM 5.3 models" (commit c8f744a), which added:
  • apps/api/src/server/models/index.ts: glm-5p3 → label "GLM 5.3", glm-5p3-flash → label "GLM 5.3 Flash" (provider zai, hostProvider fireworks_ai, requiresApiKey: false, freePlanEligible: true)
  • apps/litellm/litellm-config.json: routing entries for both models
  • apps/api/src/services/tokenUsage.ts: pricing entries for fireworks_ai/glm-5p3
  • packages/agent-runtime/src/runtime/harness/{fx,opencode,pi}/definition.ts: added both models to the Opencode and Pi harness allow-lists
  • Tests: apps/api/tests/models.test.ts, apps/api/tests/piModelsConfig.test.ts
• Both models are treated identically to the existing GLM 5.2 (No BYOK required, free-plan eligible, no internal-only flag), matching the existing docs convention for that row.

Confidence

High-confidence: this is an exact addition of two production model IDs/labels to an existing, actively-maintained reference table and harness list, directly supported by the merged implementation and test assertions. No wording, positioning, or restructuring judgment calls were needed — the new rows follow the exact format of the adjacent GLM 5.2 row.

cc @benja

Want tembo to make any changes? Add a comment with @tembo and i'll get back to work!

      

---

Note

Low Risk
Documentation-only updates to model lists and pricing tables with no runtime or security impact.

Overview
Documents GLM 5.3 (glm-5p3) and GLM 5.3 Flash (glm-5p3-flash) so public references match the production registry.

The supported-models table in features/models.mdx gains two Z.ai rows (same BYOK convention as GLM 5.2). Opencode and Pi harness model lists in features/agents.mdx now include both names. resources/pricing.mdx adds Tembo Gateway rates: 5.3 mirrors GLM 5.2 pricing; 5.3 Flash uses lower input/output/cache-read rates.

^{Reviewed by Cursor Bugbot for commit 4bd0c96. Bugbot is set up for automated code reviews on this repo. Configure here.}

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
tembo	🟢 Ready	View Preview	Aug 30, 2026, 4:37 PM

💡 Tip: Enable Workflows to automatically generate PRs for you.

[tembo]

Requesting review from @DarrenBaldwin07 who has experience with the following files modified in this PR:

• features/models.mdx

[tembo]

The model names, IDs, provider, BYOK status, and current OpenCode/Pi mappings match monorepo PR #10871. However, this is blocked by the release requirement.

• Product PR #10871 merged as c8f744a on August 30, after the latest published release v0.2.828 from August 26.
• The commit is contained in draft v0.2.829, but that release is not published and has no successful production release deployment yet. Passing PR CI and preview deployments are not production-release evidence.

Before this merges after the release, please also complete the related docs:

• Add glm-5p3 and glm-5p3-flash rows to resources/pricing.mdx; #10871 added explicit metering prices, while the pricing table currently stops at glm-5p2.
• #10871 enables both models for FX as well as OpenCode and Pi. Coordinate with open docs PR docs: update agent harnesses for fx support #719: after its FX accordion lands, the FX model list also needs these two entries.

Recommendation: keep #727 blocked until a published release containing c8f744a completes production deployment, then rebase, add the missing pricing/FX coverage, re-verify, and merge.

[tembo]

Follow-up commit: this branch's original diff added GLM 5.3 / GLM 5.3 Flash to the supported models table and the Opencode/Pi harness lists, but missed the Tembo Gateway pricing table in resources/pricing.mdx. Added glm-5p3 and glm-5p3-flash rows there, computed from the CUSTOM_MODEL_PRICES entries for these models in apps/api/src/services/tokenUsage.ts (monorepo PR #10871) using the same 1.15x markup applied to every other row in the table (verified against glm-5p2, kimi-k3, and the Anthropic rows).

cc @benja