This PR adds a new `needs_llm_token_refresh` helper method for checking
if the LLM token needs to be refreshed.
We were duplicating the check for the `x-zed-expired-token` header in a
number of spots, and it will be gaining an additional case soon.
Release Notes:
- N/A
- **title_bar: Extract platform_title_bar from title_bar**
- **file_finder no longer depends on search and agent_servers no longer
depend on language_models**
Release Notes:
- N/A
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
The rate limiter's semaphore guard was being held for the entire
duration of a turn, including during tool execution. This caused
deadlocks when subagents tried to acquire permits while parent requests
were waiting for them to complete.
## The Problem
In `run_turn_internal`, the stream (which contains the `RateLimitGuard`
holding the semaphore permit) was kept alive throughout the entire loop
iteration - including during **tool execution**:
1. Parent request acquires permit
2. Parent starts streaming, consumes response
3. Parent starts executing tools (subagents)
4. **Stream/guard still held** while tools execute
5. Subagents try to acquire permits → blocked because parent still holds
permit
6. Deadlock if all permits are held by parents waiting for subagent
children
## The Fix
Two changes were made:
1. **Drop the stream early**: Added an explicit `drop(events)` after the
stream is fully consumed but before tool execution begins. This releases
the rate limit permit so subagents can acquire it.
2. **Removed the `bypass_rate_limit` workaround**: Since the root cause
is now fixed, the bypass mechanism is no longer needed.
Note: no release notes because subagents are still feature-flagged, and
this rate limiting change isn't actually observable without them.
Release Notes:
- N/A
Users had trouble signing in due to us relying on the Copilot::global
being set, which was never the case. We've decided to use a dedicated
LSP instance just for handling auth of Copilot Chat and other goodies.
That instance is subscribed to by local Copilot instances for projects.
When the Auth instance changes it's state, local instances are prompted
to re-check their own sign in status.
Closes#47352
Co-authored-by: dino <dinojoaocosta@gmail.com>
Release Notes:
- Fixed authentication issues with Copilot.
---------
Co-authored-by: dino <dinojoaocosta@gmail.com>
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
This PR adds a thinking toggle for controlling whether to use thinking
for a model in the Zed provider:
<img width="645" height="142" alt="Screenshot 2026-01-22 at 12 34 01 PM"
src="https://github.com/user-attachments/assets/9aa543fe-e708-4840-8b38-1a6fbcb78388"
/>
Previously we would create separate "Thinking" variants of the models
that supported thinking in the model selector.
This only applies to Anthropic models in the Zed provider, currently.
This is gated behind the `cloud-thinking-toggle` feature flag.
Release Notes:
- N/A
---------
Co-authored-by: Neel <neel@zed.dev>
This PR removes the `open-ai-responses-api` feature flag and makes it so
all OpenAI requests to the Zed provider use the Responses API.
We've been running this in Nightly/Preview for a week now without any
issues.
Closes CLO-104.
Release Notes:
- N/A
Support for Kimi K2 Thinking
Release Notes:
- Added support for thinking traces when using OpenAI-API-compatible AI providers
---------
Co-authored-by: Bennet Bo Fenner <bennet@zed.dev>
## Problem
When subagents use the `edit_file` tool, it creates an `EditAgent` that
makes its own model request to get the edit instructions. These "nested"
requests compete with the parent subagent conversation requests for rate
limiter permits.
The rate limiter uses a semaphore with a limit of 4 concurrent requests
per model instance. When multiple subagents run in parallel:
1. 3 subagents each hold 1 permit for their ongoing conversation streams
(3 permits used)
2. When all 3 try to use `edit_file` simultaneously, their edit agents
need permits too
3. Only 1 edit agent can get the 4th permit; the other 2 block waiting
4. The blocked edit agents can't complete, so their parent subagent
conversations can't complete
5. The parent conversations hold their permits, so the blocked edit
agents stay blocked
6. **Deadlock**
## Solution
Added a `bypass_rate_limit` field to `LanguageModelRequest`. When set to
`true`, the request skips the rate limiter semaphore entirely. The
`EditAgent` sets this flag because its requests are already "part of" a
rate-limited parent request.
(No release notes because subagents are still feature-flagged.)
Release Notes:
- N/A
---------
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
This PR removes the code for the legacy plans.
No more users will be on this plan as of January 17th, so it's fine to
land these changes now (as they won't be released until the 21st).
Closes CLO-76.
Release Notes:
- N/A
This feature cost $15.
Up -> Tokens we're sending to the model
Down -> Tokens we've received from the model.
<img width="377" height="69" alt="Screenshot 2026-01-14 at 12 31 01 PM"
src="https://github.com/user-attachments/assets/fc15824f-de5d-466b-8cc1-329f3c1940bb"
/>
Release Notes:
- Changed the display of tokens for OpenAI models to reflect the
input/output limits.
---------
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
- **copilot: Fix double lease panic when signing out**
- **Extract copilot_chat into a separate crate**
- **Do not use re-exports from copilot**
- **Use new SignIn API**
- **Extract copilot_ui out of copilot**
Closes#7501
Release Notes:
- Fixed Copilot providing suggestions from different Zed windows.
- Copilot edit predictions now support jumping to unresolved
diagnostics.
This PR makes it so we use a proper type for the Responses API `input`
rather than a `serde_json::Value`.
It should have never used `serde_json::Value` to begin with.
Release Notes:
- N/A
## Motivation
This PR unifies the async execution infrastructure between GPUI and
other components that depend on the `scheduler` crate (such as our cloud
codebase). By having a scheduler that lives independently of GPUI, we
can enable deterministic testing across the entire stack - testing GPUI
applications alongside cloud services with a single, unified scheduler.
## Summary
This PR completes the integration of the `scheduler` crate into GPUI,
unifying async execution and enabling deterministic testing of GPUI
combined with other components that depend on the scheduler crate.
## Key Changes
### Scheduler Integration (Phases 1-5, previously completed)
- `TestDispatcher` now delegates to `TestScheduler` for timing, clock,
RNG, and task scheduling
- `PlatformScheduler` implements the `Scheduler` trait for production
use
- GPUI executors wrap scheduler executors, selecting `TestScheduler` or
`PlatformScheduler` based on environment
- Unified blocking logic via `Scheduler::block()`
### Dead Code Cleanup
- Deleted orphaned `crates/gpui/src/platform/platform_scheduler.rs`
(older incompatible version)
## Intentional Removals
### `spawn_labeled` and `deprioritize` removed
The `TaskLabel` system (`spawn_labeled`, `deprioritize`) was removed
during this integration. It was only used in a few places for test
ordering control.
cc @maxbrunsfeld @as-cii - The new priority-weighted scheduling in
`TestScheduler` provides similar functionality through
`Priority::High/Medium/Low`. If `deprioritize` is important for specific
test scenarios, we could add it back to the scheduler crate. Let me know
if this is blocking anything.
### `start_waiting` / `finish_waiting` debug methods removed
Replaced by `TracingWaker` in `TestScheduler` - run tests with
`PENDING_TRACES=1` to see backtraces of pending futures when parking is
forbidden.
### Realtime Priority removed
The realtime priority feature was unused in the codebase. I'd prefer to
reintroduce it when we have an actual use case, as the implementation
(bounded channel with capacity 1) could potentially block the main
thread. Having a real use case will help us validate the design.
## Testing
- All GPUI tests pass
- All scheduler tests pass
- Clippy clean
## Architecture
```
┌─────────────────────────────────────────────────────────────┐
│ GPUI │
│ ┌──────────────────────┐ ┌────────────────────────────┐ │
│ │ gpui::Background- │ │ gpui::ForegroundExecutor │ │
│ │ Executor │ │ - wraps scheduler:: │ │
│ │ - scheduler: Arc< │ │ ForegroundExecutor │ │
│ │ dyn Scheduler> │ └────────────┬───────────────┘ │
│ └──────────┬───────────┘ │ │
│ │ │ │
│ └──────────┬──────────────────┘ │
│ ▼ │
│ ┌───────────────────────┐ │
│ │ Arc<dyn Scheduler> │ │
│ └───────────┬───────────┘ │
│ ┌──────────────┴──────────────┐ │
│ ▼ ▼ │
│ ┌──────────────────┐ ┌────────────────────┐ │
│ │ PlatformScheduler│ │ TestScheduler │ │
│ │ (production) │ │ (deterministic) │ │
│ └──────────────────┘ └────────────────────┘ │
└─────────────────────────────────────────────────────────────┘
```
Release Notes:
- N/A
---------
Co-authored-by: Antonio Scandurra <me@as-cii.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Yara <git@yara.blue>
Co-authored-by: Zed Zippy <234243425+zed-zippy[bot]@users.noreply.github.com>
This PR adds support for using the OpenAI Responses API through the Zed
provider.
This is gated behind the `open-ai-responses-api` feature flag.
Part of CLO-34.
Release Notes:
- N/A
Add support for OpenAI's /responses endpoint for models that don't
support /chat/completions API. This enables compatibility with newer
model variants (`gpt-5-codex`, `gpt-5-pro`, `o3-pro`, etc) while
maintaining compatibility with existing configs
Changes:
- Add `supports_chat_completions` flag to model capabilities that
defaults to true for existing behavior
- Implement responses API client with streaming support as per [OpenAI
documentation](https://app.stainless.com/api/spec/documented/openai/openapi.documented.yml).
- Add `ResponseEventMapper` to convert responses events to completion
events for maintainer simplicity
- Update UI to allow toggling `chat_completions` capability
- Add `gpt-5-codex` model
Closes#38858
Release Notes:
- Added support for `gpt-5-codex` model
---------
Co-authored-by: Bennet Bo Fenner <bennet@zed.dev>
Closes #ISSUE
Problem:
- The status bar’s pending keystroke indicator (shown next to --NORMAL--
in Vim mode) didn’t clear when focus moved to another context, e.g.
hitting g in the editor then clicking the Git panel. The keymap state
correctly canceled the prefix, but observers that render the indicator
never received a “pending input changed” notification, so the UI kept
showing stale prefixes until a new keystroke occurred.
Fix:
- The change introduces a `pending_input_changed_queued` flag and a new
helper `notify_pending_input_if_needed` which will flushes the queued
notification as soon as we have an App context. The
`pending_input_changed` now resets the flag after notifying subscribers.
Before:
https://github.com/user-attachments/assets/7bec4c34-acbf-42bd-b0d1-88df5ff099aa
After:
https://github.com/user-attachments/assets/2264dc93-3405-4d63-ad8f-50ada6733ae7
Release Notes:
- Fixed: pending keybinding prefixes on the status bar now clear
immediately when focus moves to another panel or UI context.
---------
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Conrad Irwin <conrad.irwin@gmail.com>
We recently added this `InlineCode` component but I'd forgotten that
many months ago I also introduced an `inline_code` method to the Label
component which does the same thing. That means we don't need a
standalone component at all!
Release Notes:
- N/A
First up: I'm sorry if this is a low quality PR, or if this feature
isn't wanted. I implemented this because I'd like to have this
behaviour. If you don't think that this is useful, feel free to close
the PR without comment. :)
My idea is this: I love to pull random models with Ollama to try them.
At the same time, not all of them are useful for coding, or some won't
work out of the box with the context_length set. So, I'd like to change
Zed's behaviour to not show me all models Ollama has, but to limit it to
the ones that I configure manually.
What I did is add an `auto_discover` field to the settings. The idea is
that you can write a config like this:
```json
"language_models": {
"ollama": {
"api_url": "http://localhost:11434",
"auto_discover": false,
"available_models": [
{
"name": "qwen3:4b",
"display_name": "Qwen3 4B 32K",
"max_tokens": 32768,
"supports_tools": true,
"supports_thinking": true,
"supports_images": true
}
]
}
}
```
The `auto_discover: false` means that Zed won't pick up or show the
language models that Ollama knows about, and will only show me the one I
manually configured in `available_models`. That way, I can pull random
models with Ollama, but in Zed I can only see the ones that I know work
(because I've configured them).
The default for `auto_discover` (when it is not explicitly set) is
`true`, meaning that the existing behaviour is preserved, and this is
not a breaking change for configurations.
Release Notes:
- ollama: Added `auto_discover` setting to optionally limit visible
models to only those manually configured in `available_models`
- Edit prediction providers can now be configured through the settings
UI
- Cleaned up the status bar menu to only show _configured_ providers
- Added to the status bar icon button tooltip the name of the active
provider
- Only display the data collection functionality under "Privacy" for the
Zed models
- Moved the Codestral edit prediction provider out of the Mistral
section in the agent panel into the settings UI
- Refined and improved UI and states for configuring GitHub Copilot as
both an agent and edit prediction provider
#### Todos before merge:
- [x] UI: Unify with settings UI style and tidy it all up
- [x] Unify Copilot modal `impl`s to use separate window
- [x] Remove stop light icons from GitHub modal
- [x] Make dismiss events work on GitHub modal
- [ ] Investigate workarounds to tell if Copilot authenticated even when
LSP not running
Release Notes:
- settings_ui: Added a section for configuring edit prediction providers
under AI > Edit Predictions, including Codestral and GitHub Copilot.
Once you've updated you can use the following link to open it:
zed://settings/edit_predictions.providers
---------
Co-authored-by: Ben Kunkle <ben@zed.dev>
## Closes#43887
## Release Notes:
### Problem
DeepSeek's reasoning mode API requires `reasoning_content` to be
included in assistant messages that precede tool calls. Without it, the
API returns a 400 error:
```
Missing `reasoning_content` field in the assistant message at message index 2
```
### Added/Fixed/Improved
- Add `reasoning_content` field to `RequestMessage::Assistant` in
`crates/deepseek/src/deepseek.rs`
- Accumulate thinking content from `MessageContent::Thinking` and attach
it to the next assistant/tool-call message
- Wire reasoning content through the language model provider in
`crates/language_models/src/provider/deepseek.rs`
### Testing
- Verified with DeepSeek Reasoner model using tool calls
- Confirmed reasoning content is properly included in API requests
Fixes tool-call errors when using DeepSeek's reasoning mode.
---------
Co-authored-by: Ben Brandt <benjamin.j.brandt@gmail.com>
Closes#43598
Release Notes:
- bedrock: Added opt-in `allow_global` which enables global endpoints
- bedrock: Updated cross-region-inference endpoint and model list
- bedrock: Fixed Opus 4.5 access on Bedrock, now only accessible through the `allow_global` setting
This PR simplifies error and event handling by removing the
`Ok(LanguageModelCompletionEvent::Status(CompletionRequestStatus::Failed)))`
state from the stream returned by `LanguageModel::stream_completion()`,
by changing it into an `Err(LanguageModelCompletionError)`. This was
done by collapsing the valid `CompletionRequestStatus` values into
`LanguageModelCompletionEvent`.
Release Notes:
- N/A
---------
Co-authored-by: Michael Benfield <mbenfield@zed.dev>
This fixes various issues where rustfmt failed to format code due to too
long strings, most of which I stumbled across over the last week and
some additonal ones I searched for whilst fixing the others.
Release Notes:
- N/A
Automatically retry the agent's LLM completion requests when the
provider returns 429 Too Many Requests. Uses the Retry-After header to
determine the retry delay if it is available.
Many providers are frequently overloaded or have low rate limits. These
providers are essentially unusable without automatic retries.
Tested with Cerebras configured via openai_compatible.
Related: #31531
Release Notes:
- Added automatic retries for OpenAI-compatible LLM providers
---------
Co-authored-by: Bennet Bo Fenner <bennetbo@gmx.de>
Closes#42303
Ollama added tool call identifiers
(https://github.com/ollama/ollama/pull/12956) in its latest version
[v0.12.10](https://github.com/ollama/ollama/releases/tag/v0.12.10). This
broke our json schema and made all tool calls fail.
This PR fixes the schema and uses the Ollama provided tool call
identifier when available. We remain backwards compatible and still use
our own identifier with older versions of Ollama. I added a `TODO` to
remove the `Option` around the new field when most users have updated
their installations to v0.12.10 or above.
Note to reviewer: The fix to this issue should likely get cherry-picked
into the next release, since Ollama becomes unusable as an agent without
it.
Release Notes:
- Fixed tool calling when using the latest version of Ollama
This PR adds a new component to the `language_models` crate called
`ConfiguredApiCard`:
<img width="500" height="420" alt="Screenshot 2025-11-09 at 2 07@2x"
src="https://github.com/user-attachments/assets/655ea941-2df8-4489-a4da-bba34acf33a9"
/>
We were previously recreating this component from scratch with regular
divs in all LLM providers render function, which was redundant as they
all essentially looked the same and didn't have any major variations
aside from labels. We can clean up a bunch of similar code with this
change, which is cool!
Release Notes:
- N/A
Marshall Bowers