Skip to content

[Workers AI] Update model reasoning efforts #33541

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
mchenco merged 12 commits into from
Sep 24, 2026
Merged

[Workers AI] Update model reasoning efforts #33541

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
b4b30a2
[Workers AI] Correct model reasoning controls and schemas
KastanDay Sep 19, 2026
cf75fb4
[Workers AI] Document Qwen and Nemotron reasoning controls
KastanDay Sep 21, 2026
6551b4c
[Workers AI] Format model reasoning metadata
KastanDay Sep 21, 2026
ff8a201
[Workers AI] Sync reasoning metadata for model cards
KastanDay Sep 21, 2026
a405713
[Workers AI] Display max reasoning effort concisely
KastanDay Sep 21, 2026
112440c
[Workers AI] Render reasoning efforts as code
KastanDay Sep 21, 2026
e70e855
[Workers AI] Standardize model info table rows
KastanDay Sep 21, 2026
4b2a6c5
[Workers AI] Move Nemotron reasoning guidance
KastanDay Sep 21, 2026
93bae2d
[Workers AI] Publish Nemotron native reasoning controls
KastanDay Sep 22, 2026
aaa9804
[Workers AI] Regenerate reasoning model schemas from Config API
KastanDay Sep 22, 2026
27bccce
[Workers AI] Show toggle-only reasoning as Yes
KastanDay Sep 22, 2026
22ff226
[Workers AI] Update Nemotron force_nonempty_content description
KastanDay Sep 23, 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
59 changes: 57 additions & 2 deletions src/components/models/ModelDetailPage.astro
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -144,7 +144,11 @@ const authorName = model.authorName;
const requiresWorkersPaid =
model.properties.require_workers_paid === true ||
model.properties.require_workers_paid === "true";
const hasReasoningEffortGuidance = model.name === "@cf/zai-org/glm-5.2";
const hasGlm52ReasoningEffortGuidance = model.name === "@cf/zai-org/glm-5.2";
const hasNemotronReasoningEffortGuidance =
model.name === "@cf/nvidia/nemotron-3-120b-a12b";
const hasReasoningEffortGuidance =
hasGlm52ReasoningEffortGuidance || hasNemotronReasoningEffortGuidance;

// `hasSchema`: input must have at least one key.
const hasSchema = Object.keys(model.schema.input).length > 0;
Expand Down Expand Up @@ -345,7 +349,7 @@ const headings: MarkdownHeading[] = [

<ModelFeatures model={model} />

{hasReasoningEffortGuidance && (
{hasGlm52ReasoningEffortGuidance && (
<Fragment>
<AnchorHeading
depth={2}
Expand Down Expand Up @@ -379,6 +383,57 @@ const headings: MarkdownHeading[] = [
</Fragment>
)}

{hasNemotronReasoningEffortGuidance && (
<Fragment>
<AnchorHeading
depth={2}
title="Reasoning effort"
slug="reasoning-effort"
/>
<p>
Nemotron supports normal reasoning, low reasoning, and reasoning turned
off. It does not currently select these modes from the top-level
<code>reasoning_effort</code> field. Pass the corresponding options
through <code>chat_template_kwargs</code> instead.
</p>
<p>
To use low reasoning, set both <code>enable_thinking</code> and
<code>low_effort</code>:
</p>
<Code
code={`response = client.chat.completions.create(
model="@cf/nvidia/nemotron-3-120b-a12b",
messages=[{"role": "user", "content": "What is the capital of Japan?"}],
max_tokens=16000,
temperature=1.0,
top_p=0.95,
extra_body={
"chat_template_kwargs": {
"enable_thinking": True,
"low_effort": True,
}
},
)`}
lang="python"
/>
<p>
The low-effort option appends <code>{`{reasoning effort: low}`}</code>{" "}
to the latest user message. To turn reasoning off, set
<code>enable_thinking</code> to <code>False</code>. For coding agents,
set
<code>force_nonempty_content</code> to <code>True</code> in the same
<code>chat_template_kwargs</code> object.
</p>
<p>
For more information, refer to{" "}
<a href="https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8#api-client">
NVIDIA's Nemotron API client example
</a>
.
</p>
</Fragment>
)}

{
/* Defensive dead branch (model not in slice): nova-3 transport pricing. */
model.name === "@cf/deepgram/nova-3" && (
Expand Down
29 changes: 25 additions & 4 deletions src/components/models/ModelFeatures.astro
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,11 @@
* beta → Batch → request formats → partner → realtime → unit price →
* dashboard pricing link.
*/
import { modelCurrencyFormatter, type ModelView } from "~/util/models";
import {
formatModelReasoning,
modelCurrencyFormatter,
type ModelView,
} from "~/util/models";

interface Props {
model: ModelView;
Expand Down Expand Up @@ -44,6 +48,10 @@ const hasProps = Object.keys(properties).length > 0;
const price = Array.isArray(properties.price)
? (properties.price as { price: number; unit: string }[])
: null;
const reasoning = formatModelReasoning(
properties.reasoning_effort,
properties.reasoning,
);
---

{hasProps && (
Expand All @@ -54,7 +62,7 @@ const price = Array.isArray(properties.price)
<th />
</tr>
</thead>
<tbody>
<tbody data-row-stripes="off">
{properties.planned_deprecation_date && (
<tr>
<td>
Expand Down Expand Up @@ -127,10 +135,23 @@ const price = Array.isArray(properties.price)
<td>Yes</td>
</tr>
)}
{properties.reasoning && (
{reasoning && (
<tr>
<td>Reasoning</td>
<td>{properties.reasoning_effort ?? "Yes"}</td>
<td>
{Array.isArray(reasoning) ? (
<span class="flex flex-wrap gap-x-4 gap-y-1">
{reasoning.map((effort) => (
<span>
<code>{effort.value}</code>
{effort.isDefault && " (default)"}
</span>
))}
</span>
) : (
reasoning
)}
</td>
</tr>
)}
{properties.vision && (
Expand Down
62 changes: 45 additions & 17 deletions src/content/workers-ai-models/deepseek-v4-flash-0731.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,26 @@
"property_id": "reasoning",
"value": "true"
},
{
"property_id": "reasoning_effort",
"value": {
"supported_efforts": [
"max",
"high",
"low",
"none"
],
"default_effort": "high",
"default_enabled": true,
"mandatory": false,
"normalizes_to": {
"null": "high",
"minimal": "low",
"medium": "high",
"xhigh": "high"
}
}
},
{
"property_id": "price",
"value": [
Expand Down Expand Up @@ -299,24 +319,26 @@
{
"type": "string",
"enum": [
"max",
"high",
"low",
"medium",
"high"
"none"
]
},
{
"type": "null"
}
],
"description": "Constrains effort on reasoning for reasoning models (o1, o3-mini, etc.)."
"description": "Reasoning effort. Supported levels: max, high, low, none. Compatibility aliases: minimal maps to low; medium maps to high; xhigh maps to high.",
"default": "high"
},
"chat_template_kwargs": {
"type": "object",
"properties": {
"enable_thinking": {
"type": "boolean",
"default": true,
"description": "Whether to enable reasoning, enabled by default."
"description": "Whether to enable reasoning for this model."
},
"clear_thinking": {
"type": "boolean",
Expand Down Expand Up @@ -1546,24 +1568,26 @@
{
"type": "string",
"enum": [
"max",
"high",
"low",
"medium",
"high"
"none"
]
},
{
"type": "null"
}
],
"description": "Constrains effort on reasoning for reasoning models (o1, o3-mini, etc.)."
"description": "Reasoning effort. Supported levels: max, high, low, none. Compatibility aliases: minimal maps to low; medium maps to high; xhigh maps to high.",
"default": "high"
},
"chat_template_kwargs": {
"type": "object",
"properties": {
"enable_thinking": {
"type": "boolean",
"default": true,
"description": "Whether to enable reasoning, enabled by default."
"description": "Whether to enable reasoning for this model."
},
"clear_thinking": {
"type": "boolean",
Expand Down Expand Up @@ -2363,24 +2387,26 @@
{
"type": "string",
"enum": [
"max",
"high",
"low",
"medium",
"high"
"none"
]
},
{
"type": "null"
}
],
"description": "Constrains effort on reasoning for reasoning models (o1, o3-mini, etc.)."
"description": "Reasoning effort. Supported levels: max, high, low, none. Compatibility aliases: minimal maps to low; medium maps to high; xhigh maps to high.",
"default": "high"
},
"chat_template_kwargs": {
"type": "object",
"properties": {
"enable_thinking": {
"type": "boolean",
"default": true,
"description": "Whether to enable reasoning, enabled by default."
"description": "Whether to enable reasoning for this model."
},
"clear_thinking": {
"type": "boolean",
Expand Down Expand Up @@ -3610,24 +3636,26 @@
{
"type": "string",
"enum": [
"max",
"high",
"low",
"medium",
"high"
"none"
]
},
{
"type": "null"
}
],
"description": "Constrains effort on reasoning for reasoning models (o1, o3-mini, etc.)."
"description": "Reasoning effort. Supported levels: max, high, low, none. Compatibility aliases: minimal maps to low; medium maps to high; xhigh maps to high.",
"default": "high"
},
"chat_template_kwargs": {
"type": "object",
"properties": {
"enable_thinking": {
"type": "boolean",
"default": true,
"description": "Whether to enable reasoning, enabled by default."
"description": "Whether to enable reasoning for this model."
},
"clear_thinking": {
"type": "boolean",
Expand Down Expand Up @@ -4678,4 +4706,4 @@
}
},
"deprecated": false
}
}
Loading
Loading