command-code 1.19.1 → 1.21.0

package/CHANGELOG.md

@@ -1,11 +1,21 @@
 # command-code
 
+## 1.21.0
+
+- feat: add grok 4.6
+
+## 1.20.0
+
+### Minor Changes
+
+- feat: deepseek v4 pro latest (0813)
+
 ## 1.19.1
 
 ### Patch Changes
 
-- fix: Gemini models can call tools again — drop the nullable union on shell_output's `wait` (Vertex rejects `any_of` with sibling keys) and send the tool name on every tool result (Gemini requires `function_response.name`)
-- fix: OpenAI models can use deferred tools again — the loader is now `search_tools`, since the AI SDK's OpenAI Responses adapter reserves `tool_search` for its own built-in and dropped the `arguments` field, 400-ing every request after a tool lookup. The old name still works: it stays registered (hidden) so it clears the unknown-tool check, resolves through the tool-name aliases, and is rewritten on replay so it never reaches the wire. Hooks matching `TOOL_SEARCH` keep firing.
+- fix: Gemini models tool calling issues
+- fix: deferred tools on OpenAI models
 - feat: shed unnecessary tools from headless runs
 
 ## 1.19.0

package/dist/bundled/command-code-knowledge/reference/models.md

@@ -8,7 +8,7 @@ The model catalog — every id `/model`, `--model`, `model:effort` shorthand, an
 
 | Id (use EXACTLY this) | Name | Context | Efforts | $/1M in/out · cache read | Min plan | Best for |
 |---|---|---|---|---|---|---|
-| `deepseek/deepseek-v4-pro` | DeepSeek V4 Pro | 1M | high, max | $0.435/$0.87 · cache $0.003625 | Go and above | hybrid-attention long-context reasoning |
+| `deepseek/deepseek-v4-pro` | DeepSeek V4 Pro (latest) | 1M | high, max | $0.435/$0.87 · cache $0.003625 | Go and above | hybrid-attention long-context reasoning |
 | `deepseek/deepseek-v4-flash` | DeepSeek V4 Flash (latest) | 1M | high, max | $0.14/$0.28 · cache $0.0028 | Go and above | fast hybrid-attention reasoning |
 | `moonshotai/Kimi-K3` | Kimi K3 | 1M | — | $3/$15 · cache $0.3 | Go and above | long-horizon coding & knowledge work with 1M context |
 | `moonshotai/Kimi-K2.7-Code` | Kimi K2.7 Code | 256K | — | $0.95/$4 · cache $0.19 | Go and above | improved long-horizon coding with vision |
@@ -84,5 +84,6 @@ The model catalog — every id `/model`, `--model`, `model:effort` shorthand, an
 | Id (use EXACTLY this) | Name | Context | Efforts | $/1M in/out · cache read | Min plan | Best for |
 |---|---|---|---|---|---|---|
 | `xai/grok-4.5` | Grok 4.5 | 500K | low, medium, high | $2/$6 · cache $0.5 | Go and above | smartest model for coding, agentic tasks, knowledge work |
+| `xai/grok-4.6` | Grok 4.6 | 500K | low, medium, high, xhigh | $2/$6 · cache $0.5 | GOAT and above | frontier performance on coding, knowledge work, and STEM |
 
 Rates are the advertised price list resolved from the billing source of truth (promos are already baked in): input / output / cache-read per 1M tokens, plus cache-write where the provider bills it. Min plan is the cheapest public plan that serves the model, ordered Go < GOAT < Pro < Max (see plans.md). Models without an effort column entry decide their own reasoning depth. Full pricing detail: https://commandcode.ai/docs/resources/pricing-limits

package/dist/bundled/command-code-knowledge/reference/product-help.md

@@ -296,6 +296,7 @@ Valid model ids (the /model catalog — use these EXACT ids):
 - google/gemini-3.1-flash-lite
 - sakana/fugu-ultra
 - xai/grok-4.5
+- xai/grok-4.6
 - meta/muse-spark-1.1
 - meta/muse-spark-1.2
 - meta/muse-spark-1.2-contributor