MODEL GUIDE · MODEL FACTS CHECKED 2026-09-10

MonkeyCode supports model choice at the platform layer.

The public project currently names GLM, Kimi, MiniMax, Qwen, and DeepSeek. Exact model versions, endpoints, pricing, regions, limits, and availability can change and must be checked in the deployment you evaluate.

Documented model families

A current README list—not a permanent compatibility guarantee.

The names below reflect the public project README checked on 2026-10-20. “Supported” should not be interpreted as every version, region, account type, or deployment mode.

01

GLM

Listed by the current MonkeyCode project README. Verify the specific model identifier and provider route available to your installation.

02

Kimi

Listed by the current MonkeyCode project README. Verify the specific model identifier and provider route available to your installation.

03

MiniMax

Listed by the current MonkeyCode project README. Verify the specific model identifier and provider route available to your installation.

04

Qwen

Listed by the current MonkeyCode project README. Verify the specific model identifier and provider route available to your installation.

05

DeepSeek

Listed by the current MonkeyCode project README. Verify the specific model identifier and provider route available to your installation.

Primary source: MonkeyCode project README ↗

Verify the route

Turn “supported” into a checked fact in your own release.

Do the steps once per release before you depend on a model. Each step produces a check you can keep, not a claim you have to trust.

  1. STEP 1
    Select the model in this release

    Open the model picker in the release you evaluate and confirm the exact id and version. The README family name is not the selectable id.

  2. STEP 2
    Run one bounded smoke task

    Use the fixed first-task example, not a free-form prompt: starter pack and reference solution. Confirm a task is accepted end to end.

  3. STEP 3
    Record the model the call actually returned

    Capture the model id returned by the request (logs or response). Only this proves what your account used; the picker label can differ.

  4. STEP 4
    Confirm credentials and routing

    Check who owns the key, how it is scoped and rotated, and whether traffic goes direct, through a gateway, or to a local endpoint.

SymptomLikely causeWhere to look
Request rejected with an “insufficient credit / quota” style errorOne of several possible causes: the daily token allowance for the current plan is used up, the monthly credit allowance is used up, or a provider-side account quota applies. Daily tokens, monthly credits and provider quotas are separate counters, and a rejected request does not by itself identify which one was hit.Plan caps and billing period: pricing & TCO page. Actual usage and credit balance for the current period: the product account panel after signing in. Provider-side limits: the provider account dashboard.
Request rejected with a rate-limit (429) errorPer-key or per-model request limitAdd backoff/retry, raise the account limit, or spread load across keys
Model not found / unsupported in this regionId not offered for this release, region, or account typeThe picker in this release; the provider's model list for the region
Works in one deployment, fails in anotherDifferent route, key scope, or network egressCompare request paths, gateway config, and egress policy between the two

Model facts checked 2026-09-10 · page content updated 2026-10-09. Source: provider model documentation ↗.

Selection matrix

Verify the route, not only the model name.

Two deployments can select the same model family and still have different privacy, reliability, latency, and cost characteristics.

DecisionQuestion to answerEvidence to retain
AvailabilityWhich exact model IDs and versions are selectable in this release and region?Configuration screenshot and successful test request.
RoutingDoes traffic go directly to a provider, through a private gateway, or to a locally operated endpoint?DNS, proxy, firewall, and request-path evidence.
CredentialsWho owns the account and how are keys scoped, stored, rotated, and revoked?Secret inventory and rotation test.
Data handlingWhat source context is sent and what are the provider's retention and training terms?Current contract, policy, and request sample.
OperationsWhat are the rate limits, fallback behavior, observability, and outage procedure?Load test, failure test, and runbook.
Cost and qualityWhich model meets acceptance criteria at an acceptable total task cost?Same-task pilot results, including retries and review.

Model evaluation

Test with the same repositories and acceptance criteria.

A model leaderboard does not measure your environment, tools, policies, or reviewer effort.

  1. STEP 1
    Select bounded tasks

    Use representative bug fixes, tests, documentation, and small features with written acceptance criteria.

  2. STEP 2
    Hold the workflow constant

    Use the same repository state, environment, tools, prompts, and retry policy.

  3. STEP 3
    Measure accepted outcomes

    Record pass rate, human corrections, latency, tokens, model cost, compute, and review time.

  4. STEP 4
    Test failure behavior

    Exercise quota errors, timeouts, unavailable models, credential revocation, and fallback routes.

Common questions

Models and routing, answered.

Source-backed direct answers: can it use local models?, can it run offline?

Can I use DeepSeek V4.1 Flash for free?
Yes. MonkeyCode includes DeepSeek V4.1 Flash with free access, subject to your account’s current usage limits. Direct use of the DeepSeek API is a separate service with its own billing.
Which AI models does MonkeyCode support?
The public project README currently lists GLM, Kimi, MiniMax, Qwen, and DeepSeek. Exact model identifiers, versions, regions, limits, and availability can change, so confirm what is selectable in the release you evaluate.
Can MonkeyCode use local or private models?
The self-hosting guide describes the model path as an external API, a private gateway, or a locally operated model, so private or in-network models are a documented option. Confirm which model IDs and endpoints your release actually supports.
Can MonkeyCode run models fully offline?
The project documents offline deployment, which requires the model to run on an in-network endpoint rather than an external API. Verify the offline model path against your specific release and hardware.
Does the model choice change data privacy?
Yes. Two deployments can select the same model family and still differ in routing, retention, latency, and cost. Map where inference traffic goes and what the provider retains before trusting a route with sensitive code.
How should I compare models for my codebase?
Run the same bounded tasks with an identical workflow and acceptance criteria, and measure accepted outcomes, human corrections, latency, token and compute cost, and failure behavior — not a public leaderboard.
MODEL FACTS CHANGE

Confirm current availability before procurement or rollout.

Cookie settings

We use cookies only for analytics (GA4 + Matomo) to improve the docs. No ads, no tracking across sites.