AI Models for Text, Data Analysis, and Content

Frontier models (API)

C

Claude (Anthropic)

Opus 5.5, Sonnet 5.5 ($2/$10) and Fable 5.1; Haiku 4.5 for volume

Claude is Anthropic's model family in four tiers: Haiku 4.5 for volume, Sonnet 5.5 for daily work, Opus 5.5 for long-running agents and coding, and Fable 5.1 for the hardest multi-step work. Sonnet 5.5 (09/28/2026, API claude-sonnet-5-5) costs the same as Sonnet 5: $2/$10 per million tokens, with cache reads at $0.20 and cache writes at $2.50.

Haiku 4.5: $1/$5 per 1M tok.; Sonnet 5.5 and Sonnet 5: $2/$10 (Sonnet 5.5: cache reads $0.20, cache writes $2.50); Opus 5.5: $4/$20, 5-minute cache writes $5, 1-hour cache writes $8, cache reads $0.20; Opus 5: $5/$25; Fable 5.1 and Fable 5: $10/$50; Batch API -50%. Cursor Fast for Opus 5.5: $8/$40. Details →

Frontier models (API)

AI

OpenAI (GPT)

GPT-6 Astra (frontier); GPT-6.1 Sol ($2/$10, Work and Codex); GPT-5.6 Terra/Luna for volume

GPT-6.1 Sol (gpt-6.1-sol) costs $2/$10 per 1M tokens, one fifth of Astra's rate.

GPT-6.1 Sol API: $2/$10 per 1M tok. (cache read $0.10, cache write $2.50; above 272k input, 2x input and cache, 1.5x output); GPT-6 Astra API: $10/$50 (cache read $1, cache write $12.50; Fast mode 2x); GPT-5.6: Sol $4/$20 (promo through 11/21/2026; list $5/$30), Terra $2/$12, Luna $0.20/$1.20 (from 07/30/2026). Details →

Frontier models (API)

G

Google Gemini

3.8 Flash (GA), 4 Argon Fairwind-only, 3.8 Live and 3.1 Pro preview

Gemini 3.8 Flash (09/02/2026) is Google's current model for coding, agents and reasoning: 1M context, native multimodality (text, audio, image, video).

3.8 Flash: intro $0.75/$3.75 per 1M tok. through 12/31/2026, then $1.50/$7.50; 3.5 Flash-Lite $0.30/$2.50; 3.1 Pro (preview) depends on context (about $2/$12 up to 200k). Gemini 4 Argon: introductory $2/$10, then $4/$20; Fairwind only for now. Details →

Frontier models (API)

X

Grok (xAI)

Grok 4.7 (code/agents), 4.3 (1M ctx), real-time X data

Grok is the xAI model family.

Pay-as-you-go: Grok-4.7 and 4.6 $2/$6 (above 200k prompt: $4/$12, cache $0.50); Grok-4.5 $2/$6 (cache $0.30); Grok-4.3 $1.25/$2.50. Grok-4.1 Fast retired 05/15/2026. Details →

Office ecosystems

Co

Microsoft Copilot

GPT-6 Astra and Claude Opus 5.5 in GitHub Copilot; GPT-5.6 in Microsoft Copilot

Microsoft Copilot (formerly Microsoft 365 Copilot) is the assistant in Word, Excel, PowerPoint, Outlook and Teams at $30/user/mo (annual).

Microsoft Copilot: $30/user/mo (annual); Copilot Business $21, promo $18 through 12/31/2026; Copilot Chat included with M365; Copilot Studio: $200 per 25,000 Copilot Credits/mo or pay-as-you-go; GitHub Copilot: Free, Pro $10, Pro+ $39, Max $100, Business $19, Enterprise $39 (AI Credits, 1 credit = $0.01). Details →

Research & search

P

Perplexity

search with citations, Computer agent, Agent API replaces Sonar on 09/27/2026

Perplexity pairs a search engine with an LLM: every answer carries inline citations and a source list.

Free / Pro: $20/mo ($200/yr) / Max: $200/mo ($2,000/yr, 10,000 Computer credits/mo); Enterprise Pro: $40/seat/mo ($400/yr); Enterprise Max: $325/seat/mo ($3,250/yr); Agent API: tokens at provider rates + web search $0.0025/call, sandbox $0.03/session; Search API $5 per 1,000 requests. Details →

Open-weight & self-host

M

Mistral AI

Medium 3.5 (Modified MIT) and Small 4 (Apache 2.0); EU regional endpoints; GPAI Code

Mistral Medium 3.5 (04/28/2026) is the 128B dense flagship under a Modified MIT license: 256k context, text and image input, configurable reasoning; it replaced Medium 3.1, Magistral and Devstral 2. Mistral Small 4 (03/16/2026, Apache 2.0) is a 119B MoE with 6.5B active params, also 256k.

API per 1M tokens: Medium 3.5 $1.50 in / $7.50 out; Large 3 $0.50 / $1.50; Small 4 $0.15 / $0.60; Ministral 3 8B $0.15 / $0.15; Codestral $0.30 / $0.90; batch -50%. Le Chat: Free, Pro $14.99/mo, Team $24.99/user/mo, Enterprise custom. Self-hosting weights: GPU cost. Details →

Open-weight & self-host

DS

DeepSeek

V4-Pro-0813 GA, V4.1 Flash; peak/off-peak

DeepSeek V4-Pro-0813 (GA 08/13/2026) and DeepSeek-V4.1-Flash (`deepseek-flash`) are the current API snapshots: 1M context, 384K max output, thinking and non-thinking, DSA, MIT license.

Peak/off-peak since 08/16/2026 16:00 UTC: Pro off-peak $0.66/$1.98, peak $1.32/$3.96; V4.1 Flash off-peak about $0.15/$0.60 (check current pricing; cache hit $0.0028 / $0.003625). MIT open-weight = GPU cost. Details →

Open-weight & self-host

Qw

Qwen (Alibaba Cloud)

Qwen3.8-Max (API $2/$6, snapshot 0902) + Flash; open 2.4T and Qwen3.5/3.6

Qwen3.8-Max (GA 08/03/2026) is Alibaba Cloud's flagship: MoE ~2.4T / ~95B active, 1M context, multimodality (text+image+video), API $2/$6 per 1M tok.

Qwen3.8-Max API: $2/$6 per 1M tokens (implicit cache $0.25); Qwen3.7-Max $2.50/$7.50; Qwen3.7-Plus from $0.40/$1.60; Qwen3.5-Plus from $0.40/$2.40; open-weight = GPU. Details →

Open-weight & self-host

GL

GLM (Z.ai)

GLM-5.3 (weights + API); GLM-5.3-Flash (MIT); GLM-5.2 MIT

GLM is a model family from Z.ai, formerly known as Zhipu AI.

Z.ai pricing: GLM-5.3 $1.40/$4.40; GLM-5.3-Flash $0.15/$0.50; GLM-5.2 $1.40 input, $0.26 cached input and $4.40 output per 1M tokens; GLM-4.7-Flash is listed as free in the API. Self-hosting adds GPU, power and operating costs. Details →

Open-weight & self-host

Ll

Llama (Meta)

Llama 4 Scout/Maverick (04/2025), no successor; Meta frontier is Muse Spark 1.3, Llama API shut down

Llama 4 Scout (17B active, 16 experts, 109B, 10M context) and Maverick (17B/128E, 400B, 1M) from 04/05/2025 remain the last Llama generation.

Llama and Muse Glimmer weights are free (cost = GPU). Hosted on Together: Scout $0.18/$0.59, Maverick $0.27/$0.85, Llama 3.3 70B $1.04/$1.04 per 1M tokens; Groq: Llama only in the Enterprise tier (custom pricing). Meta's Llama API shut down 07/06/2026; Meta Model API serves Muse Spark 1.3 (and older), not Llama: $1.25/$4.25, Contributor tier $0.10/$0.20. Details →

Open-weight & self-host

K

Kimi (Moonshot AI)

K3 flagship, 1M ctx, open MoE; self-host from 07/27/2026

Kimi K3 (released 07/16/2026) is Moonshot AI's flagship model: approx.

K3 API: $0.30/1M cache-hit input, $3.00 cache-miss input, $15.00 output; K2.7 Code and K2.6 - $0.95 input / $0.16-0.19 cached / $4.00 output; K2.5 legacy - $0.60 / $3.00. Details →

Specialist & agentic

Ma

Manus

autonomous AI agent; independent again since 09/2026 after the unwound Meta deal

Manus (launched 03/2025, Butterfly Effect, Singapore) is an AI agent that plans and executes multi-step tasks in a cloud computer: research, reports, slides, websites, code and data analysis. Meta's acquisition (12/2025, about $2B) was blocked by China's NDRC on 04/27/2026.

Free: 300 daily credits; Pro: from $20/mo (4,000 credits) and from $40/mo (8,000 credits), higher tiers per manus.im pricing; Team: from $20 per seat/mo; annual billing 17% cheaper. Details →

Specialist & agentic

Fu

Sakana Fugu

multi-model orchestration through one API; Fugu and Fugu Ultra

Sakana Fugu, commercially released on June 22, 2026, is not a conventional standalone LLM.

Subscriptions: $20 / $100 / $200 monthly; Fugu Ultra PAYG: $5 input, $30 output and $0.50 cached input per 1M tokens, with higher rates above 272K context. Details →

Polish language models

B

Bielik (SpeakLeash)

best Polish open-weight quality, Apache 2.0

Bielik is a family of Polish language models developed by SpeakLeash and ACK Cyfronet AGH on Polish HPC infrastructure (Helios, Athena).

Open-weight (Apache 2.0): cost = GPU infrastructure; available on Hugging Face. Details →

Polish language models

PL

PLLuM

Polish 4B-70B family; base, instruct and chat models

PLLuM (Polish Large Language Model) is a model family developed by a consortium of Polish research and public institutions.

Public model weights; cost depends on model size, GPU hosting and operations. Check the license of the exact repository before deployment. Details →