List @cf/cloudflare/clef and @cf/cloudflare/clef-flash under
cloudflare-workers-ai in the classifier models table and add changelog
entries crediting #10316 and #10322.
Move the codemode script reference, including the models API, into
docs/codemode.md and keep only one line per global in the tool
description. Declared tools get a one-line note on how scripts call them,
and the codemode system prompt guidance and MCP server section are shorter.
With the default tools a GPT-5.6 request drops from about 5,300 to 3,300
tokens.
Errors now point the way back: unknown tools and models members name close
matches, models.classify() and models.generateImages() validate their
arguments, unknown or mistyped models point to getAvailableOfType(),
store() overflow explains the store, and unshown generated images get a note.
Scripts run image models with the session's credentials, like
models.classify(). Results are base64 image blocks that image() attaches
to the codemode result; usage counts toward the session cost.
Codemode sums the usage of models.classify() calls into its tool result and shows each call's cost. Usage of tools called through ctx.executeTool() is added to the calling tool's result usage instead of being dropped.
Vercel evaluation models are generated from the AI Gateway catalog and
routed through its TypeSafe-compatible /typesafe/v1/systemone endpoint.
OpenCode Zen's jev-1.13 and jev-1.13-free are routed through
/zen/v1/systemone.
Entries of only +name/-name modify the inherited tool selection, so
"defaultTools": ["+codemode"] enables codemode without repeating the
defaults. Project entries layer on top of user settings. Document how to
enable codemode without MCP and how to use classifier models from it.
This change makes system prompt text and tool changes part of the transcript rather than silently rewriting its starting conditions. This lets Pi record when instructions changed or tools became available, restore that state after resuming or navigating branches, and preserve cached prompt prefixes where the upstream supports it.
#7638 hardcoded vLLM's thinking_token_budget. Map thinkingBudgets through
thinkingTokenBudgetField, keep the boolean as a vLLM alias, clamp so the
answer keeps 1024 tokens, and document it.
fixes#8274
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Mario Zechner <badlogicgames@gmail.com>
Add the constrained sampling API and provider plumbing for OpenAI and Anthropic tool calls, including grammar sampling variants and model capability metadata.
The OpenRouterRouting interface only had only and order, but the runtime
passes the entire object as-is to the OpenRouter API's provider field.
This means all other fields worked at runtime but had no TypeScript
type-checking, no autocomplete, and no config validation.
This adds all fields from the OpenRouter provider selection API
(https://openrouter.ai/docs/guides/routing/provider-selection):
allow_fallbacks, require_parameters, data_collection, zdr,
enforce_distillable_text, ignore, quantizations, sort, max_price,
preferred_min_throughput, preferred_max_latency.
Also fixes the models.md documentation which had an invalid fallbacks
field in the example (not part of the OpenRouter API), replaces it with
a comprehensive example showing all available fields, and clarifies that
openRouterRouting is sent as-is in the provider field.
- Add minimal example showing only required fields for local models
- Rename 'Basic Example' to 'Full Example' showing all fields
- Add 'Default' column to model configuration table
- Clarify that apiKey is required but ignored by Ollama
