Skip to content

fix(model-gateway): allow the betas the pinned Claude and OpenCode CLIs send for Claude models #115

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
EricWvi merged 2 commits into from
Oct 21, 2026
Merged

fix(model-gateway): allow the betas the pinned Claude and OpenCode CLIs send for Claude models #115

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b917ad0
fix(model-gateway): allow the per-turn-control beta the pinned Claude…
ObsisMc Oct 20, 2026
b4c5e23
fix(model-gateway): allow the betas OpenCode sends for Claude models
ObsisMc Oct 20, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 4 additions & 1 deletion docs/models/model-gateway.en.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -81,7 +81,10 @@ browser-access marker never reach the supplier.

The same pinned CLI also declares `structured-outputs-2025-12-15` and
`thinking-display-updates-2026-08-18` when queries are created through Agent SDK 0.3.303 /
ACP adapter 0.94.0. Both are included in the bounded whitelist. An upgrade must verify its
ACP adapter 0.94.0, and `per-turn-control-2026-07-01` on its first request for
`claude-haiku-5-5`. The pinned OpenCode 1.19.3 declares `thinking-binding-controls-2026-08-01`
and `structured-outputs-2025-11-13` when it requests a Claude model through
`@ai-sdk/anthropic`. All of these are included in the bounded whitelist. An upgrade must verify its
observed protocol surface before extending the policy.

| Method and path | Query | Behavior |
Expand Down
4 changes: 3 additions & 1 deletion docs/models/model-gateway.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,9 @@ Claude Code 2.1.303 的 Messages 兼容面如下；允许的 beta 标记由
客户端诊断头、会话标识及 browser-access 标记不会转发到供应商。

同一固定 CLI 通过 Agent SDK 0.3.303 / ACP adapter 0.94.0 创建查询时还会声明
`structured-outputs-2025-12-15` 与 `thinking-display-updates-2026-08-18`；它们同样在受限白名单中。
`structured-outputs-2025-12-15` 与 `thinking-display-updates-2026-08-18`，对 `claude-haiku-5-5` 的首个请求还会带
`per-turn-control-2026-07-01`。固定的 OpenCode 1.19.3 经 `@ai-sdk/anthropic` 请求 Claude 模型时声明
`thinking-binding-controls-2026-08-01` 与 `structured-outputs-2025-11-13`。它们同样在受限白名单中。
升级版本仍须先验证实际协议面，不能按客户端传入内容自动放开。

| 方法与路径 | 查询 | 行为 |
Expand Down
8 changes: 6 additions & 2 deletions internal/modelgateway/anthropic.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,8 +10,9 @@ import (

const anthropicVersion = "2023-06-01"

// These protocol capabilities were observed with the pinned Claude 2.1.303 CLI.
// A CLI upgrade must exercise its requests before extending this set.
// These protocol capabilities were observed with the pinned Claude 2.1.303 CLI and, for
// Claude models over Anthropic Messages, the pinned OpenCode 1.19.3 CLI. A CLI upgrade must
// exercise its requests before extending this set.
var claudeBetaHeaders = map[string]bool{
"claude-code-20250219": true,
"interleaved-thinking-2025-05-14": true,
Expand All @@ -22,6 +23,9 @@ var claudeBetaHeaders = map[string]bool{
"effort-2025-11-24": true,
"structured-outputs-2025-12-15": true,
"thinking-display-updates-2026-08-18": true,
"per-turn-control-2026-07-01": true,
"thinking-binding-controls-2026-08-01": true,
"structured-outputs-2025-11-13": true,
}

// runtimeModelRoute accepts only inference and the frozen binding's local model listing.
Expand Down
50 changes: 29 additions & 21 deletions internal/modelgateway/anthropic_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,27 +16,35 @@ import (

const pinnedClaudeBetas = "claude-code-20250219,interleaved-thinking-2025-05-14,thinking-token-count-2026-05-13,context-management-2025-06-27,prompt-caching-scope-2026-01-05,mid-conversation-tool-changes-2026-07-01,effort-2025-11-24"

// Agent SDK query creation enables these additional capabilities on the same pinned native CLI.
func TestPinnedClaudeAgentSDKCapabilities(t *testing.T) {
betas := pinnedClaudeBetas + ",structured-outputs-2025-12-15,thinking-display-updates-2026-08-18"
var calls atomic.Int64
service, _ := anthropicFixture(t, http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
calls.Add(1)
if r.Header.Get("Anthropic-Beta") != betas || r.URL.RawQuery != "beta=true" {
t.Error("SDK capabilities changed while forwarding")
}
w.Header().Set("Content-Type", "application/json")
_, _ = io.WriteString(w, "{\"id\":\"m\",\"type\":\"message\",\"role\":\"assistant\",\"model\":\"fixture-model\",\"content\":[{\"type\":\"text\",\"text\":\"{}\"}],\"stop_reason\":\"end_turn\",\"usage\":{\"input_tokens\":3,\"output_tokens\":1}}")
}))
request := httptest.NewRequest(http.MethodPost, "/runtime/anthropic/v1/messages?beta=true", strings.NewReader("{\"model\":\"fixture-model\",\"messages\":[],\"max_tokens\":16}"))
request.Header.Set("Authorization", "Bearer "+strings.Repeat("r", 40))
request.Header.Set("Content-Type", "application/json")
request.Header.Set("Anthropic-Version", anthropicVersion)
request.Header.Set("Anthropic-Beta", betas)
response := httptest.NewRecorder()
service.ModelHandler(response, request)
if response.Code != 200 || calls.Load() != 1 {
t.Fatalf("pinned SDK request rejected: status=%d calls=%d", response.Code, calls.Load())
// Each pinned CLI's exact first-request header for a Claude model is forwarded unchanged.
func TestPinnedCLIClaudeModelBetas(t *testing.T) {
for name, betas := range map[string]string{
// Claude Code 2.1.303 through the pinned ACP adapter, with claude-haiku-5-5.
"claude": "interleaved-thinking-2025-05-14,thinking-token-count-2026-05-13,context-management-2025-06-27,prompt-caching-scope-2026-01-05,claude-code-20250219,per-turn-control-2026-07-01,mid-conversation-tool-changes-2026-07-01,effort-2025-11-24",
// OpenCode 1.19.3 through @ai-sdk/anthropic, with claude-haiku-5-5.
"opencode": "thinking-binding-controls-2026-08-01,structured-outputs-2025-11-13",
} {
t.Run(name, func(t *testing.T) {
var calls atomic.Int64
service, _ := anthropicFixture(t, http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
calls.Add(1)
if r.Header.Get("Anthropic-Beta") != betas {
t.Error("pinned betas changed while forwarding")
}
w.Header().Set("Content-Type", "application/json")
_, _ = io.WriteString(w, "{\"id\":\"m\",\"type\":\"message\",\"role\":\"assistant\",\"model\":\"fixture-model\",\"content\":[{\"type\":\"text\",\"text\":\"{}\"}],\"stop_reason\":\"end_turn\",\"usage\":{\"input_tokens\":3,\"output_tokens\":1}}")
}))
request := httptest.NewRequest(http.MethodPost, "/runtime/anthropic/v1/messages", strings.NewReader("{\"model\":\"fixture-model\",\"messages\":[],\"max_tokens\":16}"))
request.Header.Set("X-Api-Key", strings.Repeat("r", 40))
request.Header.Set("Content-Type", "application/json")
request.Header.Set("Anthropic-Version", anthropicVersion)
request.Header.Set("Anthropic-Beta", betas)
response := httptest.NewRecorder()
service.ModelHandler(response, request)
if response.Code != 200 || calls.Load() != 1 {
t.Fatalf("pinned %s betas rejected: status=%d calls=%d", name, response.Code, calls.Load())
}
})
}
}

Expand Down
Loading