server: answer reasoning_effort levels the template rejects with 400 (#32332)

The reasoning effort probe already renders every candidate level, but
it dropped the ones the template raises on, so a request with such a
level (e.g. max on Qwen3.8) reached the template and came back as a
500 on /v1/chat/completions, /v1/responses and /v1/messages.

Keep the raising levels in reasoning_effort_rejected and reject them
in oaicompat_chat_params_parse with an invalid_argument, which the
server answers with a 400 listing the supported levels. Aliases and
levels a template ignores are not rejected, so high on Qwen3.8
still renders as xhigh.

Also trim the output_config.effort test to a single level.
This commit is contained in:
Pascal authored and GitHub committed 2026-11-16 06:55:56 +01:00
1 parent 05db09a4d3
commit a946eb541a
8 files changed
+54 -4

No files matched your search

+1
View File
@@ -1979,5 +1979,6 @@ common_json common_chat_templates_get_caps(const common_chat_templates * chat_te
: chat_templates->template_default->caps;
common_json result = caps.to_map();
result["reasoning_effort_levels"] = caps.reasoning_effort_levels;
result["reasoning_effort_rejected"] = caps.reasoning_effort_rejected;
return result;
}
+11 -1
View File
@@ -627,9 +627,14 @@ caps caps_get(jinja::program & prog) {
// levels sharing a render are aliases and the highest one is kept
const auto unknown = render("REASONING_EFFORT_UNKNOWN");
std::vector<std::string> renders;
std::vector<std::string> rejected;
for (const auto & level : candidates) {
const auto rendered = render(level);
if (!rendered || rendered == unknown) {
if (!rendered) {
rejected.push_back(level);
continue;
}
if (rendered == unknown) {
continue;
}
const auto it = std::find(renders.begin(), renders.end(), *rendered);
@@ -640,6 +645,11 @@ caps caps_get(jinja::program & prog) {
result.reasoning_effort_levels.push_back(level);
}
}
// a template that renders no level fails the probe for another reason, so its failures say nothing about the levels
if (!result.reasoning_effort_levels.empty()) {
result.reasoning_effort_rejected = rejected;
}
}
JJ_DEBUG("%s\n", result.to_string().c_str());
+3
View File
@@ -23,6 +23,9 @@ struct caps {
// reasoning effort levels the template renders distinctly, from lowest to highest
std::vector<std::string> reasoning_effort_levels;
// reasoning effort levels the template raises on
std::vector<std::string> reasoning_effort_rejected;
// one of the 2 content capabilities must be true
bool supports_string_content = true;
bool supports_typed_content = false;
+11
View File
@@ -8614,6 +8614,17 @@ static void test_reasoning_effort_caps() {
assert_effort_levels("models/templates/openai-gpt-oss-120b.jinja", all_levels);
assert_effort_levels("models/templates/Cohere2MoE.jinja", {});
assert_effort_levels("models/templates/Qwen-Qwen3-0.6B.jinja", {});
auto assert_effort_rejected = [](const std::string & path, const std::vector<std::string> & expected) {
auto tmpls = read_templates(path);
assert_equals(json(expected).dump(), common_chat_templates_get_caps(tmpls.get()).at("reasoning_effort_rejected").dump());
};
assert_effort_rejected("models/templates/IFM-K2-Horizon.jinja", { "minimal", "xhigh", "max" });
assert_effort_rejected("models/templates/tencent-Hy3.jinja", { "minimal", "medium", "xhigh", "max" });
assert_effort_rejected("models/templates/deepseek-ai-DeepSeek-V4.jinja", {});
assert_effort_rejected("models/templates/openai-gpt-oss-120b.jinja", {});
assert_effort_rejected("models/templates/Cohere2MoE.jinja", {});
}
static void test_msg_diffs_compute() {
+1 -1
View File
@@ -1031,7 +1031,7 @@ By default, it is read-only. To make POST request to change global properties, y
- `voices` - the predefined voices of an audio generation model, to pass as `voice` to `/tts`. Only present when the model has some (e.g. LFM3-Audio, Pocket TTS without voice cloning)
- `model_path` - the path to model file (same with `-m` argument)
- `chat_template` - the model's original Jinja2 prompt template
- `chat_template_caps` - capabilities of the chat template (see `common/jinja/caps.h` for more info), `reasoning_effort_levels` lists the `reasoning_effort` values the template renders distinctly, from lowest to highest
- `chat_template_caps` - capabilities of the chat template (see `common/jinja/caps.h` for more info), `reasoning_effort_levels` lists the `reasoning_effort` values the template renders distinctly, from lowest to highest, and `reasoning_effort_rejected` the ones it raises on, which are answered with a 400
- `modalities` - the list of supported modalities
- `is_sleeping` - sleeping status, see [Sleeping on idle](#sleeping-on-idle)
+5
View File
@@ -1646,6 +1646,11 @@ json oaicompat_chat_params_parse(
inputs.enable_thinking = false;
inputs.chat_template_kwargs.erase("reasoning_effort");
} else if (!reasoning_effort.empty()) {
const auto rejected = caps.at("reasoning_effort_rejected").get<std::vector<std::string>>();
if (std::find(rejected.begin(), rejected.end(), reasoning_effort) != rejected.end()) {
const auto levels = caps.at("reasoning_effort_levels").get<std::vector<std::string>>();
throw std::invalid_argument("reasoning_effort \"" + reasoning_effort + "\" is not supported by the chat template, supported levels: " + string_join(levels, ", "));
}
inputs.chat_template_kwargs["reasoning_effort"] = json(reasoning_effort).dump();
}
}
@@ -851,13 +851,13 @@ def test_anthropic_thinking():
assert res.body["type"] == "message"
@pytest.mark.parametrize("effort", ["low", "high"])
def test_anthropic_output_config_effort(effort):
def test_anthropic_output_config_effort():
"""output_config.effort must reach the chat template, same as chat_template_kwargs.reasoning_effort"""
global server
server.jinja = True
server.chat_template_file = '../../../models/templates/openai-gpt-oss-120b.jinja'
server.start()
effort = "low"
def count_tokens(extra):
res = server.make_request("POST", "/v1/messages/count_tokens", data={
+20
View File
@@ -23,6 +23,26 @@ def create_server():
server.n_slots = 1
def test_reasoning_effort_rejected_by_template():
global server
server.jinja = True
# the template raises on max, and high is one of its levels
server.chat_template_file = '../../../models/templates/IFM-K2-Horizon.jinja'
server.start()
messages = [{"role": "user", "content": "Hello"}]
for path, data in [
("/chat/completions/input_tokens", {"messages": messages, "reasoning_effort": "max"}),
("/v1/responses/input_tokens", {"input": "Hello", "reasoning": {"effort": "max"}}),
("/v1/messages/count_tokens", {"model": "test", "messages": messages, "output_config": {"effort": "max"}}),
]:
res = server.make_request("POST", path, data=data)
assert res.status_code == 400, f"{path}: {res.body}"
res = server.make_request("POST", "/chat/completions/input_tokens", data={"messages": messages, "reasoning_effort": "high"})
assert res.status_code == 200, res.body
@pytest.mark.parametrize("tools", [None, [], [TEST_TOOL]])
@pytest.mark.parametrize("template_name,reasoning,expected_end", [
("deepseek-ai-DeepSeek-R1-Distill-Qwen-32B", "on", "<think>\n"),