Fast AI models: 73 low-latency options for chat and autocomplete

Prices from official vendor price lists · October 2026
73 small, fast models for high-volume work where response time matters more than depth. Output prices run from $0.063 to $12 per 1M tokens across the 56 models with a paid per-token price; 32 have a free tier and 27 ship open weights.
Low latency
Services found: 73
#AI modelCapabilitiesAPI $/1M (in / out)Context windowPrice / moOutput $/1MOverviewVisit site
61

Aion-3.0-Mini

Aion Labs
Text & chatLow latencyMultilingual
$0.7 / $1.4131.1K tokensAPI only1.4View
62

Step 3.5 Flash

Stepfun
Text & chatVisionLow latency
$0.1 / $0.3262.1K tokensAPI only0.3View
63

Seed-2.0-Lite

Bytedance Seed
Text & chatVisionVideo
$0.25 / $2262.1K tokensAPI only2View
64

Granite 4.0 Micro

IBM
Text & chatReasoningTool calling
$0.017 / $0.112131K tokensAPI only0.112View
65

GPT-5.1-Codex-Mini

OpenAI
Text & chatVisionCoding
$0.25 / $2400K tokensAPI only2View
66

Nano Banana (Gemini 2.5 Flash Image)

Google
VisionImage generationWeb search grounding
$0.3 / $2.532.8K tokensAPI only2.5View
67

Seed 2.1 Turbo

Bytedance Seed
Text & chatVisionVideo
$0.5 / $2.5262.1K tokensAPI only2.5View
68

MiniMax M2.1

MiniMax
Text & chatReasoningTool calling
$0.3 / $1.2204.8K tokensAPI only1.2View
69

Nano Banana Pro (Gemini 3 Pro Image)

Google
VisionAdjustable thinkingImage generation
$2 / $12131.1K tokensAPI only12View
70

MiniMax M2-her

MiniMax
Text & chatReasoningTool calling
$0.3 / $1.265.5K tokensAPI only1.2View
71

MiniMax M2

MiniMax
Text & chatReasoningTool calling
$0.255 / $1.02204.8K tokensAPI only1.02View
72

GPT Audio Mini

OpenAI
Text & chatAudioStructured JSON output
$0.6 / $2.4128K tokensAPI only2.4View
73

GPT-5 Image Mini

OpenAI
VisionImage generationStructured JSON output
$2.5 / $2400K tokensAPI only2View

Browse by need

Reasoning models169 modelsCoding assistants17 modelsVision and images156 modelsLong context 1M+82 modelsFree tier available80 modelsOpen weights88 models

These 73 models are the small, fast members of each family - Flash, Mini, Lite, Nano and Haiku variants - built for workloads where response time and request volume matter more than depth. Output runs from $0.10 to $5 per million tokens, an order of magnitude below the frontier.

Typical uses: routing and classification, autocomplete, moderation passes, first-line support. GLM 5.3 Flash, Gemini 3.8 Flash and Gemini 3.7 Flash head the list. These models are weaker at long multi-step reasoning, so the common pattern is a fast model in front, escalating to a frontier model only when the task warrants it.

Small by designThese are the Flash, Mini, Lite and Haiku variants, not throttled versions of big models.
Cheap enough to overuseAt $0.10 per 1M tokens you can afford a model call where you used to write a rule.
Escalate, do not compromisePair with a frontier model for the cases the fast one cannot close.

Models with this capability

The 73 models below all support this capability. Prices are input and output per 1M tokens from the vendor official price list; for dirhams multiply by 3.6725. Tap a card to open the model page.

GLM 5.3 Flash
#1
Z.ai
in $0.075out $0.251.31M tokens
Gemini 3.8 Flash
#2
Google DeepMind
in $0.75out $3.751.05M tokens
Gemini 3.7 Flash
#3
Google DeepMind
in $0.75out $3.751.05M tokens
GPT-5.6 Luna
#4
OpenAI
in $0.2out $1.21.05M tokens
DeepSeek V4 Flash 0731
#5
DeepSeek
in $0.065out $0.181.31M tokens
Gemini 3.6 Flash
#6
Google DeepMind
in $0.75out $3.751.05M tokens
MiniMax M3
#7
MiniMax
in $0.3out $1.21.05M tokens
Inkling Small
#8
Thinkingmachines
in $0.45out $1.21.05M tokens

How to choose

  • Measure time to first token, not just total time — perceived speed lives there.
  • Test on your hardest realistic input; small models degrade suddenly rather than gradually.
  • Check context limits, which are often smaller on the fast variants.
  • Design the escalation rule before launch, not after the first complaint.

Common mistakes

Using a fast model for long reasoningIt will answer quickly and be wrong quickly.
No escalation pathEvery product has a tail of requests a small model cannot handle.

Who this suits

Real-time interfacesAutocomplete, inline suggestions, live moderation.
High-volume classificationMillions of routing decisions where each one is simple.
Cost-constrained productsFree tiers and low-ARPU features that still need model calls.

Related filters

Frontier tier41 flagship models, the best each vendor sells
Free tier80 models with a free API allowance
Batch API140 models with half-price async processing
Realtime voice7 models for live voice conversation
All 239 modelsThe full rating with filters and prices

Frequently asked questions

What do I give up for speed?

Depth on long multi-step reasoning and reliability in extended tool chains. For routing, classification and autocomplete you give up nothing that matters.

What is the common architecture?

A fast model in front, escalating to a frontier model only when confidence is low or the task is complex.

Which are the cheapest?

GLM 5.3 Flash, Gemini 3.8 Flash and Gemini 3.7 Flash, with output from $0.10 per 1M tokens.

How much faster are they really?

Typically several times, and the gap widens on short answers.

Do they support tools and JSON?

Most do, though reliability on long tool chains is lower.

Frequently searched

By capabilityReasoningCodingVisionLong context 1M+Free tierOpen weightsLow latencyRealtime voiceImage generation
By vendorAnthropic (13)OpenAI (30)Google DeepMind (17)xAI (8)Meta (6)DeepSeek (9)Mistral AI (11)Alibaba (Qwen) (34)Z.ai (12)Moonshot AI (5)
Popular modelsClaude Fable 5.1GPT-6 AstraClaude Opus 5Claude Fable 5GPT-5.6 SolGLM 5.3Grok 4.6Kimi K3GPT-5.6 Terra
By deliveryBatch APIPrompt cachingPriority tierFrontier tierProprietary APITool callingStructured JSONWeb searchMultilingual

Ratings 0

Avg. score 0

Rate this page

Help us improve