| name | dust-llm |
| description | Step-by-step guide for adding support for a new LLM in Dust. Use when adding a new model, or updating a previous one. |
Adding Support for a New LLM Model
This skill guides you through adding a newly released LLM to the model_constructors +
llms stack (the endpoint-class router). It replaces the legacy lib/api/llm/clients/*
router, which no longer exists.
Mental model
A model reaches production through three stacked layers. Add the new model to each:
- Model config (
front/types/assistant/models/*) — the legacy ModelConfigurationType
describing the model (context, vision, reasoning efforts, pricing tiers). Still the source
of truth consumed by the UI, pricing, and the dust layer.
model_constructors (front/lib/model_constructors/*) — provider-agnostic endpoint
classes, one per (provider, model, region, provider-api). Each class mixes a shared
provider base client with a per-model config mixin (input schema, context size,
token pricing). This is where the real request/response shape and the narrowed input
config live.
llms (dust layer) (front/lib/llms/*) — thin Dust-specific wrappers around the
model_constructors classes that add Dust concerns (display name, byok, endpoint
filters, and any caps — e.g. exposing 250k context on a model that natively supports
1M). Registered into DUST_STREAM_ENDPOINTS.
Endpoints are named and filed as:
{provider}_{model}_{region}_{provider_api}.ts
e.g. google_gemini_3_6_flash_global_agent_platform.ts. The class name is the
PascalCase of the same, with numbers spelled out:
GoogleGeminiThreeDotSixFlashGlobalAgentPlatformStream.
The fastest, most reliable way to add a model is to copy the most recent model in the
same family across all layers and rename. Grep every reference to that model and mirror
each one. This skill lists the reference points; the sibling model is your template.
Before you start: verify against official docs (MANDATORY)
You MUST confirm every value below against the provider's official documentation and leave a
URL + date in a code comment next to it. Do not carry values over from memory.
- Specs (context window, max output tokens, vision, structured output):
- OpenAI:
https://platform.openai.com/docs/models
- Anthropic:
https://docs.anthropic.com/en/docs/about-claude/models/overview
- Google:
https://ai.google.dev/gemini-api/docs/models
- Mistral:
https://docs.mistral.ai/getting-started/models/models_overview/
- Pricing (input / output / cached input per 1M tokens):
- OpenAI:
https://openai.com/api/pricing/
- Anthropic:
https://www.anthropic.com/pricing#anthropic-api
- Google:
https://ai.google.dev/gemini-api/docs/pricing
- Mistral:
https://mistral.ai/technology/#pricing
- Host / region availability: verify which provider APIs and regions actually serve the
model day-one. Mirror the sibling model's endpoints, but only register an endpoint whose
region is actually available. Keep an unavailable-but-anticipated endpoint class defined and
unregistered (see Gemini's EU agent-platform example) with a comment saying why.
WebSearch/WebFetch the docs first. If a value can't be confirmed, surface it — don't guess.
Reference points to mirror (grep the sibling model)
Pick the newest sibling (e.g. for "Gemini 3.6 Flash" the sibling is "Gemini 3.5 Flash") and
grep -rln its id / const / class-name / model-id string. You will touch, roughly:
A. Model config + central registry
| File | What to add |
|---|
front/types/assistant/models/{provider}.ts | X_MODEL_ID const + X_MODEL_CONFIG. Set isLatest: false on the previous model in the same family and drop "latest" from its description. |
front/types/assistant/models/models.ts | Add id to STATIC_MODEL_IDS and config to SUPPORTED_MODEL_CONFIGS (imports in both alpha blocks). |
front/types/assistant/models/auto.ts | If the model should participate in auto/auto_fast/auto_complex routing, add a ModelStreamCandidate. |
front/lib/model_constructors/types/models.ts | Add export const X = "model-id" and include it in the MODELS array (this is the model_constructors id type). |
B. Pricing / tiers / reasoning (TYPE-ENFORCED over StaticModelIdType)
Adding the id to STATIC_MODEL_IDS makes these fail to compile until updated:
| File | What to add |
|---|
front/lib/api/assistant/token_pricing/global.ts | CURRENT_MODEL_PRICING entry (input/output/cache_read_input_tokens per 1M) + doc URL comment. |
front/lib/api/assistant/token_pricing/static_model_reasoning_efforts.ts | { none, light, medium, high } support map. Must match the config's supportedReasoningEfforts (enforced by tiers.test.ts). |
front/lib/api/assistant/token_pricing/tiers.ts | STATIC_MODEL_TIERS entry mapping each supported effort → tier name. |
C. model_constructors — the endpoint classes (stream)
| File | What to add |
|---|
front/lib/model_constructors/providers/{provider}/models/{model}.ts | Config mixin WithXConfig(Base) exposing static model, static configSchema, static contextSize, static maxOutputTokens. Reuse the provider's shared inputConfig/reasoning_efforts/shared helpers. contextSize/maxOutputTokens are the REAL provider values — caps belong in the dust layer. |
front/lib/model_constructors/stream/endpoints/{provider}_{model}_{region}_{api}.ts | One class per available (region, provider-api), extending WithXConfig(BaseClient). Set static tokenPricing (per-endpoint, region-adjusted), region, regionalEndpoint, and static id = this.buildId(). Base clients live in stream/clients/*. |
front/lib/model_constructors/stream/index.ts | Import + register each available endpoint in STREAM_ENDPOINTS. |
D. model_constructors — tests (TDD, see below)
| File | What to add |
|---|
front/lib/model_constructors/test/endpoints/{...}.test.ts | One StreamSetup per endpoint. Copy the sibling's key set, but start every case at null — never copy its expected values (see the TDD loop). |
front/lib/model_constructors/test/endpoints/setups.ts | Import + register each registered endpoint's setup (satisfies Record<StreamEndpointId, StreamSetup> forces completeness). |
E. llms — the dust layer (stream)
| File | What to add |
|---|
front/lib/llms/providers/{provider}/models/{model}.ts | Dust config mixin WithDustXConfig(Base) — Object.assignes the legacy X_MODEL_CONFIG onto the class and overrides displayName/description/byok (and any caps). |
front/lib/llms/stream/endpoints/{...}.ts | One thin dust wrapper per endpoint extending the model_constructors class via the dust mixin; call defineDustStreamEndpoint(...). |
front/lib/llms/stream/index.ts | Register each available dust endpoint in DUST_STREAM_ENDPOINTS (satisfies Record<StreamEndpointId, ...>). |
F. SDK + UI + marketing mirror
| File | What to add |
|---|
sdks/js/src/types.ts | Add the id to the KnownModelLLMId union. Then rebuild the SDK types (cd sdks/js && npm run build:types) so front's sdk_drift.test.ts (which reads the built @dust-tt/client) passes. |
front/components/providers/model_configs.ts | Add config to USED_MODEL_CONFIGS so it shows in the UI. |
marketing/types/assistant/models/models.ts | Add { modelId, displayName, providerId } snapshot. |
marketing/lib/api/assistant/token_pricing.ts | Add the pricing entry (keep in sync with front). |
Batch endpoints (.../batch/...) are a curated subset — only add them if the model
needs batch. They are NOT completeness-enforced. Set supportsBatchProcessing to the real
capability regardless.
The TDD loop (steps to actually run)
The endpoint classes derive their behavior from a shared integration test harness. Let the
live API tell you the input contract — never infer it from the sibling model. Sibling
expectations are the single biggest source of wrong config: two models in the same family
routinely differ on temperature, reasoning efforts, and forced tool use.
The config schema must ALWAYS mirror the API's real behavior as closely as possible. It
describes what the provider accepts — not what Dust happens to send today, and not what would
be convenient. If the API accepts a value, the schema accepts it; if the API rejects a value,
the schema rejects it. Never narrow past the API because an upstream layer already strips the
field (the dropTemperature / dropTemperatureWhenReasoning config parsers in lib/llms are
a product policy and belong there, not in the endpoint schema), and never widen past it to
avoid a union. Concretely: Anthropic reasoning models accept exactly temperature: 1, so the
field is z.literal(1).optional().default(1) — not z.undefined(), even though the Dust layer
drops it before the endpoint ever sees it.
When a divergence from the API is genuinely wanted (exposing a narrower effort set to control
cost, say), it is a policy choice — write it as a comment stating that the API allows more
and why Dust doesn't, so the next reader doesn't mistake it for a provider constraint.
Reasoning efforts must ALWAYS mirror the model's official documentation, not merely whatever
the endpoint happens to accept. This is the one place where "what the API tolerates" is the wrong
source of truth, because gateways are routinely looser than the models they serve:
- The Fireworks gateway validates
reasoning_effort against low/medium/high/xhigh/max/none
for every model it hosts, so a live run "passes" on efforts the model never defined.
- DeepSeek documents disabled/high/max and says low/medium are mapped to high and xhigh to
max — accepting them would silently rewrite the caller's choice.
- Kimi K3 is documented low/high/max by Moonshot;
medium works through Fireworks but is not
a K3 effort.
- Kimi K2.6 has binary thinking; the graded values are accepted and do nothing (measured:
low produced more reasoning than medium).
- grok-4.5 silently accepts
minimal and xhigh, which xAI documents only for other models.
So: find the model author's doc (not just the host's), expose exactly the efforts it lists,
and link it in a comment. Where host and author docs disagree, follow the author unless the host
documents a model-specific override — generic host guidance is not a contradiction. Then confirm
each documented effort actually works on the live endpoint, and record any effort the endpoint
accepts but the docs omit, with a note that undocumented efforts can change without notice.
When the product still offers an effort the model does not have, map it in the llms layer with
a configParsers entry (mapReasoningNoneToMinimal, mapNonNoneReasoningToHigh,
mapReasoningEffortToLowHighMax, forceHighReasoningEffort) — never with a schema .transform(), and never by widening the
endpoint schema to swallow it.
1. Widen
Write the config mixin with configSchema set to the broad inputConfigSchema
(front/lib/model_constructors/types/input/configuration.ts), marked // TDD SCAFFOLD. Every
case must reach the API instead of being short-circuited by a guessed schema.
2. Write the test with every case null
Copy the sibling's key set (so coverage matches) but not its expected values. null
runs the case with its default checkers. Starting from the sibling's
INPUT_CONFIGURATION_ERROR markers hides exactly the differences you are trying to find, and
lets stale expectations survive — a suite whose expectations were never run green will happily
assert things the schema makes impossible.
3. Red run — the whole suite, no --bail
You want every failure at once in order to characterize the contract:
cd front
NODE_ENV=test RUN_LLM_TEST=true DUST_MANAGED_{PROVIDER}_API_KEY=... \
npm run test -- --config lib/model_constructors/test/vite.config.js \
lib/model_constructors/test/endpoints/{...}.test.ts
Env-var names live in the sibling's createInstance (DUST_MANAGED_ANTHROPIC_API_KEY,
DUST_MANAGED_GOOGLE_AI_STUDIO_API_KEY, …). Agent-platform/Vertex endpoints need
VERTEX_AI_PROJECT_ID plus GCP credentials — a GOOGLE_APPLICATION_CREDENTIALS service-account
key works and needs no gcloud auth application-default login. Add --bail 1 or
-t "<substring>" only later, when iterating on a single case.
4. Sort every failure into one of three buckets
The bucket decides the fix:
| Last event | Meaning | What to do |
|---|
error carrying a provider message (invalid_request_error, …) | Real API constraint | The schema must encode it |
error of type input_configuration_error | Our own zod rejected it before any request | With the widest schema this means a converter or base client still rejects it |
| The case passes | The API accepts this input | Whether to allow it is a policy choice — match the sibling unless there's a reason to diverge, and state which you chose and why |
A passing case is evidence. It disproves any assumption that the model rejects that input —
including assumptions already written down. Do not keep an INPUT_CONFIGURATION_ERROR because
a code comment says the model doesn't support something: the run outranks the comment.
5. Narrow — including the defaults
Rewrite configSchema to the real contract, with a doc URL + date in a comment next to each
value. Three things to pin deliberately, not by inheritance:
reasoning default effort — read it off the official doc, every time. The .default(...)
is load-bearing: an absent reasoning sends no thinking config, so the provider's own default
applies, and that differs per model (adaptive-on for Fable 5 / Opus 5 / Sonnet 5; thinking-off
for Opus 4.8/4.7/4.6 and Sonnet 4.6; no thinking for Haiku 4.5; max for Kimi K3 and GLM-5.2).
Never carry over a sibling's default or invent one for cost reasons — Kimi K3 sat at low
when Moonshot documents max. Mirror the documented default and cite the page; if the product
wants a cheaper default, that belongs in defaultReasoningEffort on the llms model config, not
in the endpoint schema.
temperature handling. Sweep actual values against the API rather than assuming — the
rule is per-model. Anthropic reasoning models accept only 1 while thinking is on and any
value while thinking is off; some reject the field outright.
- Effort set and
forceTool compatibility. Which efforts are genuinely accepted, and
whether a forced tool_choice may coexist with reasoning.
6. Green run
Mark the genuinely-rejected cases INPUT_CONFIGURATION_ERROR, re-run the full suite until
every case passes, then delete the // TDD SCAFFOLD comment.
7. Re-run every endpoint sharing the mixin
A config mixin is shared across regions and provider APIs (e.g. global/anthropic +
eu/agent-platform), so narrowing it changes all of them. Run each one.
8. Push the new behavior up into the family's shared config
Shared configs are per family — Opus, Sonnet, Haiku each have their own; a family with a
single member (Fable 5) just keeps a standalone config. A family's shared config should
track the latest member of that family, because the next model in it is far likelier to
repeat the newest behavior than the oldest. So when characterizing a model reveals that its
family's shared config was wrong, fix the shared config and put the override on the older
models — never special-case the newest one.
The reflex to resist is the opposite: leaving the shared config alone and giving the new model
a bespoke schema. That makes every future model in the family inherit stale behavior, and it
is how a restriction that only ever applied to one old model ends up applied to all of them.
(Worked example: forceTool: z.undefined() sat in the shared Opus config because extended
thinking forbids a forced tool_choice. Opus 4.7, 4.8 and 5 all use adaptive thinking and
all accept it — verified live — so the fix was to drop it from the shared config, not to
override it on Opus 5.)
Do not merge families that happen to agree today. Fable 5 and Opus 5 share every value except
one (Fable 5 cannot disable thinking), but they are different families, so they keep separate
configs and the coincidence is allowed to drift.
Then re-run the suites of every model in the family (§7), since they all moved.
If you cannot run the live suite (no key / non-interactive), narrow the config from the
sibling model in the same family and say so explicitly — flag every expectation as
unverified; the live run must still happen before merge.
Without NODE_ENV=test+RUN_LLM_TEST, the test file loads but its cases are skipped; that
still validates it compiles and is registered.
Verify (non-live checks that must pass)
cd front
npx tsgo --noEmit
NODE_ENV=test npm run test -- \
types/assistant/models/sdk_drift.test.ts \
types/assistant/models/types.test.ts \
lib/api/assistant/token_pricing/tiers.test.ts
tsgo clean over the files you touched (the satisfies Record<...> maps and STREAM_ENDPOINT_SETUPS are your completeness guardrails).
sdk_drift.test.ts green ⇒ front ⊆ SDK (rebuild sdks/js types if it names your id).
tiers.test.ts green ⇒ reasoning-effort maps in sync with the configs.
Model config properties (quick ref)
| Property | Notes |
|---|
contextSize / generationTokensCount | Real provider values (legacy config). Caps go in the dust layer. |
supportsVision | Can process images. |
supportsResponseFormat | Structured output (JSON). Often incompatible with tool use — verify. |
supportedReasoningEfforts | { none, light, medium, high }. Must match static_model_reasoning_efforts.ts. |
defaultReasoningEffort | Default effort. |
isLatest / isLegacy | Exactly one isLatest per family; flip the previous one to false. |
regionalAvailability | { "us-central1", "europe-west1" } — reflect real availability. |
tokenizer | Tokenizer for token counting. |
Checklist
Troubleshooting
sdk_drift.test.ts names your id → add it to KnownModelLLMId in sdks/js/src/types.ts, then cd sdks/js && npm run build:types (the test reads the built @dust-tt/client).
tsgo on setups.ts / index files → you added an endpoint to STREAM_ENDPOINTS without a matching setup, or vice-versa. Register both.
tiers.test.ts fails → static_model_reasoning_efforts.ts disagrees with the config's supportedReasoningEfforts.
- Model not in UI → missing from
USED_MODEL_CONFIGS.
- Live test rejects a config → check the bucket first (§4). A provider
invalid_request_error means narrow configSchema and mark the case INPUT_CONFIGURATION_ERROR; an input_configuration_error under the widened scaffold means a converter or base client is rejecting it, not the API.
- A case you expected to fail passes → the model accepts that input. Fix the expectation (and any comment claiming otherwise) rather than keeping the marker.
- Live suite 401s → check the key you actually exported. A shell profile can define the same
DUST_MANAGED_*_API_KEY twice; the last export wins interactively, so grepping for the first match can hand you a stale key.