Singularity Logs

THE MODEL, IN DETAIL / deepseek-ai

DeepSeek-V3

deepseek/deepseek-chatOpen weights

DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations...

OpenRouter
Source-linked model intelligenceCatalog added 26 Dec 2024Specifications observed 17 Oct 2026

GSM8K

89.3%

Source-reported

Context window

163.8K

16.4K max output tokens

Input price

$0.32

USD / million input tokens

Output price

$0.89

USD / million output tokens

Inference providers

2

3 provider routes observed

Measured output speed

—

No model-level measurement

01 / PERFORMANCE

A clearer view of capability.

Read the score, understand the context, and explore where this model sits among its peers.

Performance fingerprint

Individual published accuracy scores · not a composite index

At least three published accuracy scores are needed for this view.

In good company

GSM8K · OpenEvals / Hugging Face

89.5
Qwen
89.3
This model
86.2
microsoft
84.5
meta-llama

Rank #2 in this published cohort. 4 models in the source cohort. Showing selected nearby models. Evaluation settings can differ.

Evidence, by task.

Coverage is not a capability score.

2published results
1 evaluation source
Reasoning1 evaluation
CodingAwaiting evidence
Mathematics1 evaluation
VisionAwaiting evidence
AgenticAwaiting evidence
Long contextAwaiting evidence

Documented features, task evaluations, and preference rankings describe different things. Missing coverage does not imply a model cannot perform a task.

Explore our methodology

02 / BENCHMARK EXPLORER

The numbers behind the model.

Original results, exact versions, and a direct path to the evidence behind every score.

2 published results
020406080100GSM8K: 89.3% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face89.3%GSM8KHigher is betterMMLU-Pro: 64.4% · Published snapshot 2026-03-16 · OpenEvals / Hugging Face64.4%MMLU-ProHigher is better

2 percentage-based evaluations. Each benchmark measures a different task; compare results within the same evaluation and protocol.

Published benchmark results for DeepSeek-V3
BenchmarkScoreVersion & metricSourcePublishedEvidence
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026
Published snapshot 2026-03-16Accuracy · Higher is betterOpenEvals / Hugging FaceSource-reported15 Mar 2026

Design Arena

Source-reported Elo across individual design categories

3d1,087
models arena · Rank #89 · 50.5% win rate
Codecategories1,110
models arena · Rank #100 · 48.4% win rate
Dataviz1,090
models arena · Rank #101 · 50% win rate
Gamedev1,055
models arena · Rank #106 · 43.7% win rate
Svg976
models arena · Rank #87 · 38.8% win rate
Uicomponent1,092
models arena · Rank #94 · 52.7% win rate
Website1,122
models arena · Rank #99 · 48.5% win rate

Catalog observation: 17 Oct 2026. Category ratings have separate comparison pools.

03 / COST & VALUE

Know what your workload costs.

Explore token prices, see where the spend goes, and model a workload of your own.

Input$0.32per million tokens
Output$0.89per million tokens
Cache read—per million tokens
Cache write—per million tokens

Workload cost calculator

Change the numbers. See the trade-offs.

Cache-read pricing is not documented for this selection.

Token estimate only. Excludes cache writes, tools, and other provider fees. Actual token usage and provider limits vary.

Your estimated spend

1,000 requests · 11M tokens

$4.09estimated total
Uncached input$3.2
Cached input$0
Output$0.89
Per request$0.0041
Without cache discount$4.09
Estimated cache savings$0

04 / INFERENCE NETWORK

One model. Different ways to run it.

Compare inference routes by price, context, quantization, and reported availability.

2 providers · 3 routes
Provider routes for DeepSeek-V3, observed 17 Oct 2026
Provider / routeInput / MOutput / MCache / MContextOutput limitSpeedLatencyAvailability · 1dRates
DDeepInfraFP4 · flex$0.256$0.712—163.8K147.5K——57.11%
SStreamLakeQuantization not disclosed$0.2574$1.0287—128K16K——99.93%
DDeepInfraFP4$0.32$0.89—163.8K16.4K——95.69%

Snapshot observed 17 Oct 2026. Speed and latency are the reported 30-minute median when available. This is a dated snapshot, not live monitoring.

Provider price comparison

USD per million tokens · input and output rates

InputOutput
DeepInfrafp4 · flex
Input
$0.256
Output
$0.712
StreamLake
Input
$0.2574
Output
$1.0287
DeepInfrafp4
Input
$0.32
Output
$0.89

Reported availability

Provider availability over the selected observation window

DeepInfra · flex · FP457.11%
StreamLake99.93%
DeepInfra · FP495.69%

Output speed

Reported 30-minute p50 · tokens per second

No provider throughput measurements were returned in this snapshot.

Response latency

Reported 30-minute p50 · seconds

No provider latency measurements were returned in this snapshot.

Context by provider

Actual route limits can differ from the catalog maximum

DeepInfra · flex · FP4163.8K
StreamLake128K
DeepInfra · FP4163.8K

05 / CAPABILITIES & MODALITIES

What goes in. What comes out.

A map of documented interfaces and API features, separate from benchmark performance.

INPUT MODALITIES

Text
DeepSeek-V3Model interface

OUTPUT MODALITIES

Text

Feature support

Listed catalog API parameters and documented pricing features

ReasoningNot listed
Tool callingSupported
Tool choiceSupported
Structured outputsSupported
Response formatSupported
Reasoning detailsNot listed
Seed controlSupported
Stop sequencesSupported
Prompt cachingNot documented
TemperatureSupported

Room to work

Context capacity and completion limits

163.8Ktotal context tokens
Context window
16.4Kmax output

Output tokens share the total context budget. The maximum completion length is a separate limit, and individual provider routes may have smaller limits.

Catalog context
163,840 tokens
Maximum output
16,384 tokens
Top-provider context
163,840 tokens

Features vary by route

Check the provider before depending on a specific API feature.

API feature support by provider
Provider routeReasoningToolsStructured outputResponse formatImplicit caching
DeepInfra · flex · FP4Not listedSupportedSupportedSupportedNot listed
StreamLakeNot listedSupportedSupportedSupportedNot listed
DeepInfra · FP4Not listedSupportedSupportedSupportedNot listed

06 / UNDER THE HOOD

The complete specification sheet.

Exact identities, model architecture, limits, and the defaults that shape a request.

Identity & architecture

Developer
deepseek-ai
Exact identity
deepseek-ai/DeepSeek-V3
Subject
Model
Access
Open weights
License
Not disclosed
Parameters
Not disclosed
Tokenizer
DeepSeek
Instruction format
Not disclosed

Reasoning & controls

Reasoning
Not listed
Mandatory reasoning
Not documented
Reasoning by default
Not documented
Published configuration
Not disclosed
Supported reasoning efforts
Not documented
Default reasoning effort
Not disclosed
API parameters
15
Provider moderation
No
Quantization variants
fp4

Dates & limits

Catalog added
26 Dec 2024
Knowledge cutoff
2024-07-31
Catalog expiration
Not documented
Specifications observed
17 Oct 2026
Providers observed
17 Oct 2026
Request limits
Provider-specific / not supplied
Canonical catalog ID
deepseek/deepseek-chat-v3

API parameters & defaults

Availability is catalog-wide. Defaults can vary across providers.

frequency_penaltySupported · default not supplied
logit_biasSupported · default not supplied
max_tokensSupported · default not supplied
min_pSupported · default not supplied
presence_penaltySupported · default not supplied
repetition_penaltySupported · default not supplied
response_formatSupported · default not supplied
seedSupported · default not supplied
stopSupported · default not supplied
structured_outputsSupported · default not supplied
temperatureSupported · default not supplied
tool_choiceSupported · default not supplied
toolsSupported · default not supplied
top_kSupported · default not supplied
top_pSupported · default not supplied

07 / BUILD WITH THIS MODEL

From exploration to your first request.

Use the documented catalog identifier and keep your API key in an environment variable.

Make an API request

const response = await fetch(
  "https://openrouter.ai/api/v1/chat/completions",
  {
    method: "POST",
    headers: {
      Authorization: `Bearer ${process.env.OPENROUTER_API_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      "model": "deepseek/deepseek-chat",
      "messages": [
        {
          "role": "user",
          "content": "Explain the trade-offs in this design."
        }
      ]
    }),
  },
);
if (!response.ok) throw new Error(`HTTP ${response.status}`);
const data = await response.json();
console.log(data.choices[0].message.content);

08 / PROVENANCE & HISTORY

Every number has a story.

Follow the source, check the observation date, and distinguish published results from model specifications.

The observation trail

Recorded events, in chronological order

Added to the model catalog

Catalog creation date, as recorded by OpenRouter.

GSM8K result published

89.3% · OpenEvals / Hugging Face

MMLU-Pro result published

64.4% · OpenEvals / Hugging Face

Model specifications observed

Catalog pricing, limits, and supported parameters captured.

Provider routes observed

3 inference routes in this snapshot.

Observation dates describe when data was captured. They do not represent newly run evaluations or continuous performance monitoring.

Published observation ledger

Historical evidence, with source and version preserved.

Published observation history for DeepSeek-V3
ObservedBenchmarkScoreVersionSourceVerification
13 Oct 2026MMLU-ProPublished snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported
13 Oct 2026GSM8KPublished snapshot 2026-03-16OpenEvals / Hugging FaceSource-reported

09 / A FEW MORE DETAILS

Questions worth asking.

Quick answers, drawn from this model’s connected specifications.

How much does DeepSeek-V3 cost?

Catalog starting rates are $0.32 per million input tokens and $0.89 per million output tokens. Compare provider routes and service tiers above for their specific prices.

How much context can it handle?

163,840 tokens in the catalog context window. The catalog lists a maximum of 16,384 completion tokens. Provider limits can differ.

Does it support tools and structured output?

Tool calling: listed as supported. Structured outputs: listed as supported. See the provider feature matrix for route-specific differences.

How should I interpret these benchmark results?

Read the benchmark name, exact version, metric, source, and model configuration together. Different evaluation settings can change the result. Preference ratings and source-defined composite indices remain separate from task accuracy.

Are these prices and provider measurements live?

This page uses dated source snapshots. Catalog data was observed on 17 Oct 2026. Provider routes were observed on 17 Oct 2026. Missing measurements are not estimated or treated as zero.

KEEP EXPLORING

More from deepseek-ai.

Explore related identities from the same developer, or take them into a side-by-side comparison.

THE RIGHT MODEL STARTS WITH THE RIGHT CONTEXT.

Put the details side by side.

Explore the trade-offs in a direct model comparison.

Compare DeepSeek-V3
Source-backed evidence. Different tasks. A clearer picture.
AboutMethodologyBenchmark dashboard
DeepSeek-V3 — Singularity Logs