fix(tasks): allow GLM reasoning levels

[tatoalo]

Problem

GLM 5.2 supports high and max reasoning, but the task backend did not register the model, so valid selections were rejected before a run started. This complements PostHog/code#3615.

Changes

• Register GLM 5.2 with only its supported high and max reasoning levels.
• Extend the task API regression matrix to accept those levels and reject medium.

[greptile-apps]

Prompt To Fix All With AI

Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.

---

### Issue 1 of 1
products/tasks/backend/temporal/process_task/utils.py:99-103
**GLM Inherits Anthropic Provider Routing**

Selecting this model through `runtime_adapter="claude"` derives `provider="anthropic"`, and that provider and the `@cf/zai-org/glm-5.2` model name are passed unchanged to the agent server. If provider selects the upstream API or credentials, valid `high` and `max` requests now pass validation but fail when Anthropic receives the Cloudflare-style model identifier.

_{Reviews (1): Last reviewed commit: "fix(tasks): allow GLM reasoning levels" | Re-trigger Greptile}

[stamphog]

Note

🤖 stamphog reviewed d228ff4ba34ef3fe0b094b7819f73300d80fe08f — verdict: REFUSED

Greptile flagged an unresolved, plausible correctness issue: selecting GLM via runtime_adapter="claude" hardcodes provider="anthropic" while the model string stays the Cloudflare-style "@cf/zai-org/glm-5.2" identifier, so newly-allowed high/max requests could pass validation here but fail downstream. It's unaddressed.

• Author wrote 74% of the modified lines and has 124 merged PRs in these paths (familiarity STRONG).
• greptile-apps[bot] reviewed the current head.
• Unresolved Greptile inline comment on products/tasks/backend/temporal/process_task/utils.py:99-103 about provider/model mismatch for the new GLM entry — no reply or follow-up commit addresses it.
• Provider is derived solely from runtime_adapter (always 'anthropic' for 'claude'), with no override for the Cloudflare-namespaced GLM model id, so the mismatch Greptile describes is visible in this diff and not contradicted by anything else in the file.

Gate mechanics and policy version

Gate		Result
prerequisites	✓	all clear
deny-list	✓	no deny categories matched
size	✓	4L, 1F substantive, 54L/2F incl. docs/generated/snapshots — within ceiling
tier	✓	T1-agent / T1b-small (54L, 2F, single-area, fix)
stamphog 2.0.0b3		.stamphog/policy.yml @ 7b7de1b · reviewed head d228ff4

[deployment-status-posthog]

Deploy status

Environment	Status	Deployed At	Workflow
dev	✅ Deployed	2026-07-21 14:08 UTC	Run
prod-us	✅ Deployed	2026-07-21 14:36 UTC	Run
prod-eu	✅ Deployed	2026-07-21 14:39 UTC	Run