vndangkhoa/zed
1
0
Fork 0
mirror of https://github.com/zed-industries/zed.git synced 2026-05-31 19:05:00 +07:00
Code Issues Projects Releases Packages Wiki Activity Actions 6

opencode: Model updates (#57076)

Browse source 
**TL;DR**: clearer docs + models cleanup.

----

**Docs**: 
- as per the discussion in
https://github.com/zed-industries/zed/issues/56869, added a note to the
docs highlighting that temporary models should be configured using
Custom Models. Adding a whole example felt redundant considering the
full example is literally 2 rows below.


**Model updates**:
- **Ring 2.6 1T Free**: removed
- **GLM 5 and GLM 5.1**: different settings based on subscription —
[131k](8e710e19ea/providers/opencode/models/glm-5.1.toml (L22))
[output](8e710e19ea/providers/opencode/models/glm-5.toml (L22))
on OpenCode but
[32k](8e710e19ea/providers/opencode-go/models/glm-5.1.toml (L22))
[output](8e710e19ea/providers/opencode-go/models/glm-5.toml (L22))
on OpenCode Go
- **MiniMax M2.5**: different settings based on subscription — [131k
output on
OpenCode](8e710e19ea/providers/opencode/models/minimax-m2.5.toml (L22))
and [65k on OpenCode
Go](8e710e19ea/providers/opencode-go/models/minimax-m2.5.toml (L19))
- **Nemotron 3 Super Free**: enabled interleaved reasoning as per
[docs](8e710e19ea/providers/opencode/models/nemotron-3-super-free.toml (L13)).
Ran some quick tests and confirmed everything seems to work fine
(_"rename this variable. add a simple function. remove the function.
tell me a joke"_)
- **GPT 5.3 Codex Spark**: removed image support as per
[docs](https://github.com/anomalyco/models.dev/blob/dev/providers/opencode/models/gpt-5.3-codex-spark.toml#L23-L25)

The [docs say GLM 5 in OpenCode Zen has a deprecation date of May
14](https://opencode.ai/docs/zen/#deprecated-models) but that seems to
still be active and is [not marked as deprecated on
models.dev](8e710e19ea/providers/opencode/models/glm-5.toml)
so I didn't remove it yet 🤷

----

Self-Review Checklist:

- [x] I've reviewed my own diff for quality, security, and reliability
- [x] Unsafe blocks (if any) have justifying comments
- [x] The content is consistent with the [UI/UX
checklist](https://github.com/zed-industries/zed/blob/main/CONTRIBUTING.md#uiux-checklist)
- [ ] Tests cover the new/changed behavior
- [x] Performance impact has been considered and is acceptable

Closes https://github.com/zed-industries/zed/issues/56869

Release Notes:
- OpenCode: updated models (removed Ring 2.6 1T Free, enabled
interleaved reasoning for Nemotron 3 Super Free, deleted incorrect image
support for GPT 5.3 Codex Spark, and updated token counts for MiniMax
M2.5, GLM 5, and GLM 5.1)
- OpenCode Free: clearer docs for temporary free models
This commit is contained in:
Vlad Ionescu 2026-05-18 20:15:04 +03:00 committed by GitHub
parent b9ba43c9c1
commit 7dcd422a73
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 42 additions and 26 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
crates/language_models/src/provider/opencode.rs
View file

@ -602,11 +602,11 @@ impl LanguageModel for OpenCodeLanguageModel {
}
fn max_token_count(&self) -> u64 {
self.model.max_token_count()
self.model.max_token_count(self.subscription)
}
fn max_output_tokens(&self) -> Option<u64> {
self.model.max_output_tokens()
self.model.max_output_tokens(self.subscription)
}
fn stream_completion(
@ -646,7 +646,9 @@ impl LanguageModel for OpenCodeLanguageModel {
request,
self.model.id().to_string(),
1.0,
self.model.max_output_tokens().unwrap_or(8192),
self.model
.max_output_tokens(self.subscription)
.unwrap_or(8192),
mode,
anthropic::completion::AnthropicPromptCacheMode::Automatic,
);
@ -671,7 +673,7 @@ impl LanguageModel for OpenCodeLanguageModel {
self.model.id(),
false,
false,
self.model.max_output_tokens(),
self.model.max_output_tokens(self.subscription),
reasoning_effort,
self.model.interleaved_reasoning(),
);
@ -692,7 +694,7 @@ impl LanguageModel for OpenCodeLanguageModel {
self.model.id(),
false,
false,
self.model.max_output_tokens(),
self.model.max_output_tokens(self.subscription),
None,
supports_none_reasoning_effort,
);

54
crates/opencode/src/opencode.rs
View file

@ -141,8 +141,6 @@ pub enum Model {
MimoV2_5,
#[serde(rename = "big-pickle")]
BigPickle,
#[serde(rename = "ring-2.6-1t-free")]
Ring2_6_1TFree,
#[serde(rename = "nemotron-3-super-free")]
Nemotron3SuperFree,
#[serde(rename = "qwen3.5-plus")]
@ -204,10 +202,9 @@ impl Model {
| Self::DeepSeekV4Flash => &[OpenCodeSubscription::Go],
// Free models
Self::MiniMaxM2_5Free
| Self::Nemotron3SuperFree
| Self::BigPickle
| Self::Ring2_6_1TFree => &[OpenCodeSubscription::Free],
Self::MiniMaxM2_5Free | Self::Nemotron3SuperFree | Self::BigPickle => {
&[OpenCodeSubscription::Free]
}
// Custom models get their subscription from settings, not from here
Self::Custom { .. } => &[],
@ -263,7 +260,6 @@ impl Model {
Self::Qwen3_5Plus => "qwen3.5-plus",
Self::Qwen3_6Plus => "qwen3.6-plus",
Self::BigPickle => "big-pickle",
Self::Ring2_6_1TFree => "ring-2.6-1t-free",
Self::Nemotron3SuperFree => "nemotron-3-super-free",
Self::Custom { name, .. } => name,
@ -316,7 +312,6 @@ impl Model {
Self::Qwen3_5Plus => "Qwen3.5 Plus",
Self::Qwen3_6Plus => "Qwen3.6 Plus",
Self::BigPickle => "Big Pickle",
Self::Ring2_6_1TFree => "Ring 2.6 1T Free",
Self::Nemotron3SuperFree => "Nemotron 3 Super Free",
Self::Custom {
@ -378,7 +373,6 @@ impl Model {
| Self::DeepSeekV4Pro
| Self::DeepSeekV4Flash
| Self::BigPickle
| Self::Ring2_6_1TFree
| Self::Nemotron3SuperFree => ApiProtocol::OpenAiChat,
Self::Custom { protocol, .. } => *protocol,
@ -395,8 +389,8 @@ impl Model {
| Self::MimoV2_5Pro
| Self::Glm5
| Self::Glm5_1
| Self::BigPickle
| Self::Ring2_6_1TFree => true,
| Self::Nemotron3SuperFree
| Self::BigPickle => true,
Self::Custom {
interleaved_reasoning,
@ -407,7 +401,7 @@ impl Model {
}
}
pub fn max_token_count(&self) -> u64 {
pub fn max_token_count(&self, subscription: OpenCodeSubscription) -> u64 {
match self {
// Anthropic models
Self::ClaudeOpus4_7 => 1_000_000,
@ -436,13 +430,18 @@ impl Model {
// OpenAI-compatible models
Self::MiniMaxM2_7 => 204_800,
Self::MiniMaxM2_5 | Self::MiniMaxM2_5Free => 204_800,
Self::Glm5 | Self::Glm5_1 => 202_725,
Self::Glm5 | Self::Glm5_1 => {
if subscription == OpenCodeSubscription::Go {
202_752
} else {
204_800
}
}
Self::KimiK2_6 | Self::KimiK2_5 => 262_144,
Self::MimoV2_5Pro => 1_048_576,
Self::MimoV2_5 => 1_000_000,
Self::Qwen3_5Plus | Self::Qwen3_6Plus => 262_144,
Self::BigPickle => 200_000,
Self::Ring2_6_1TFree => 262_000,
Self::Nemotron3SuperFree => 204_800,
Self::DeepSeekV4Pro | Self::DeepSeekV4Flash => 1_000_000,
@ -450,7 +449,7 @@ impl Model {
}
}
pub fn max_output_tokens(&self) -> Option<u64> {
pub fn max_output_tokens(&self, subscription: OpenCodeSubscription) -> Option<u64> {
match self {
// Anthropic models
Self::ClaudeOpus4_7 | Self::ClaudeOpus4_6 => Some(128_000),
@ -485,10 +484,22 @@ impl Model {
// OpenAI-compatible models
Self::MiniMaxM2_7 => Some(131_072),
Self::MiniMaxM2_5 | Self::MiniMaxM2_5Free => Some(131_072),
Self::Glm5 | Self::Glm5_1 => Some(32_768),
Self::MiniMaxM2_5Free => Some(131_072),
Self::MiniMaxM2_5 => {
if subscription == OpenCodeSubscription::Go {
Some(65_536)
} else {
Some(131_072)
}
}
Self::Glm5 | Self::Glm5_1 => {
if subscription == OpenCodeSubscription::Go {
Some(32_768)
} else {
Some(131_072)
}
}
Self::BigPickle => Some(128_000),
Self::Ring2_6_1TFree => Some(66_000),
Self::KimiK2_6 | Self::KimiK2_5 => Some(65_536),
Self::Qwen3_5Plus | Self::Qwen3_6Plus => Some(65_536),
Self::DeepSeekV4Pro | Self::DeepSeekV4Flash => Some(384_000),
@ -525,7 +536,6 @@ impl Model {
| Self::Gpt5_4Mini
| Self::Gpt5_4Nano
| Self::Gpt5_3Codex
| Self::Gpt5_3Spark
| Self::Gpt5_2
| Self::Gpt5_2Codex
| Self::Gpt5_1
@ -536,6 +546,9 @@ impl Model {
| Self::Gpt5Codex
| Self::Gpt5Nano => true,
// OpenAI models without image support
Self::Gpt5_3Spark => false,
// Google models support images
Self::Gemini3_1Pro | Self::Gemini3Flash => true,
@ -556,7 +569,6 @@ impl Model {
| Self::DeepSeekV4Pro
| Self::DeepSeekV4Flash
| Self::BigPickle
| Self::Ring2_6_1TFree
| Self::Nemotron3SuperFree => false,
Self::Custom { protocol, .. } => matches!(
@ -571,7 +583,7 @@ impl Model {
pub fn supported_reasoning_effort_levels(&self) -> Option<Vec<ReasoningEffort>> {
match self {
Self::Ring2_6_1TFree | Self::MimoV2_5Pro | Self::MimoV2_5 => Some(vec![
Self::MimoV2_5Pro | Self::MimoV2_5 => Some(vec![
ReasoningEffort::Low,
ReasoningEffort::Medium,
ReasoningEffort::High,

2
docs/src/ai/llm-providers.md
View file

@ -656,6 +656,8 @@ By default, models from all subscription types are shown. Optionally, you can hi
}
```
**Note:** Zed only bundles configuration for long-term OpenCode Free models! Free models that are only available for a limited time are not included in Zed. To use such models, create a Custom Model using the configuration settings published on [the OpenCode website](https://opencode.ai/docs/zen#pricing) and on [models.dev](https://github.com/anomalyco/models.dev/tree/dev/providers/opencode/models).
#### Custom Models {#opencode-custom-models}
The Zed agent comes pre-configured with OpenCode models. If you wish to use newer models or models with custom endpoints, you can do so by adding the following to your Zed settings file ([how to edit](../configuring-zed.md#settings-files)):