Vision & Multimodal
ZICQ LLM rankings update daily across intelligence, coding, price, and context for selection and comparison.
Last updated: 56 min ago (2026-11-21 05:34)Sources and methodology
Multimodal models with image / video / document input; ranked by the Intelligence index with a vision capability bonus.
2. Benchmarks:Intelligence, coding, and agentic indices are sourced directly from Artificial Analysis (artificialanalysis.ai) official benchmark indices.
3. Overall Score Weights:Overall Score = 60% × Intelligence Index + 25% × Coding Index + 15% × Agentic Index. Models without benchmark data are ranked at the end.
4. Referral links:"Try it" buttons marked "Referral" are referral links; ZICQ may earn a commission. Referral relationships never affect rankings or scores.
Statistics for this selection
⚡ Side-by-Side Model PK & Token Cost Calculator Click "+ Compare" on any model below to swap · Live monthly cost calculator · Shareable comparison URL
| Metric | Anthropic: Claude Fable 5.2 Anthropic | Anthropic: Claude Fable 5.2 (batch) Anthropic | OpenAI: GPT-6.2 Astra OpenAI |
|---|---|---|---|
| Intelligence / Coding / Agentic | 64.6 / 0.0 / 0.0 | 64.6 / 0.0 / 0.0 | 62.1 / 0.0 / 0.0 |
| Context Window | 1M tokens | 1M tokens | 1.1M tokens |
| Input / Output Price (1M) | $8.00 / $40.00 | $4.00 / $20.00 | $8.00 / $40.00 |
| Est. Monthly Cost (30d) | Adjust inputs to calculate | Adjust inputs to calculate | Adjust inputs to calculate |
-
#401
The relace-search model uses 4-12 `view_file` and `grep` tools in parallel to explore a codebase and return relevant files to the user request. In contrast to RAG, relace-search performs agentic...
Tools StructuredContext 256K Output ≤ 128K$1.00/1M $3.00/1M outStable Image Understanding -
#402
DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...
Tools Reasoning StructuredContext 163K Output ≤ 147K$0.26/1M $0.38/1M out cache $0.13/1MStable Image Understanding -
#403
Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in...
Tools Reasoning StructuredContext 262K Output ≤ 98K$0.60/1M $2.50/1M out cache $0.15/1MStable Image Understanding -
#404
Voxtral Small is an enhancement of Mistral Small 3, incorporating state-of-the-art audio input capabilities while retaining best-in-class text performance. It excels at speech transcription, translation and audio understanding. Input audio...
Tools Audio Docs StructuredContext 32K Output ≤ 26K$0.10/1M $0.30/1M out cache $0.01/1MStable Image Understanding -
#405
gpt-oss-safeguard-20b is a safety reasoning model from OpenAI built upon gpt-oss-20b. This open-weight, 21B-parameter Mixture-of-Experts (MoE) model offers lower latency for safety tasks like content classification, LLM filtering, and trust...
Tools Reasoning StructuredContext 131K Output ≤ 65K$0.07/1M $0.30/1M out cache $0.04/1MStable Image Understanding -
#406
Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model to handle more complex...
Tools Reasoning StructuredContext 204K Output ≤ 131K$0.50/1M $2.00/1M out cache $0.10/1MStable Image Understanding -
#407
DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...
Tools Reasoning StructuredContext 163K Output ≤ 147K$0.27/1M $0.41/1M outStable Image Understanding -
#408
Uncensored and creative writing model based on Mistral Small 3.2 24B with good recall, prompt adherence, and intelligence.
StructuredContext 131K Output ≤ 117K$0.30/1M $0.50/1M out cache $0.15/1MStable Image Understanding -
#409
Relace Apply 3 is a specialized code-patching LLM that merges AI-suggested edits straight into your source files. It can apply updates from GPT-4o, Claude, and others into your files at...
Context 256K Output ≤ 128K$0.85/1M $1.25/1M outStable Image Understanding -
#410
DeepSeek-V3.1 Terminus is an update to [DeepSeek V3.1](/deepseek/deepseek-chat-v3.1) that maintains the model's original capabilities while addressing issues reported by users, including language consistency and agent capabilities, further optimizing the model's...
Tools Reasoning StructuredContext 163K Output ≤ 147K$0.27/1M $1.00/1M outStable Image Understanding -
#411
Qwen3 Coder Flash is Alibaba's fast and cost efficient version of their proprietary Qwen3 Coder Plus. It is a powerful coding agent model specializing in autonomous programming via tool calling...
Tools StructuredContext 1M Output ≤ 65K$0.20/1M $0.97/1M out cache $0.04/1MStable Image Understanding -
#412
Qwen3-Next-80B-A3B-Instruct is an instruction-tuned chat model in the Qwen3-Next series optimized for fast, stable responses without “thinking” traces. It targets complex tasks across reasoning, code generation, knowledge QA, and multilingual...
Tools StructuredContext 262K Output ≤ 235K$0.10/1M $1.10/1M out cache $0.07/1MStable Image Understanding -
#413
Kimi K2 0905 is the September update of [Kimi K2 0711](moonshotai/kimi-k2). It is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32...
Tools StructuredContext 262K Output ≤ 98K$0.60/1M $2.50/1M outStable Image Understanding -
#414
Hermes 4 is a large-scale reasoning model built on Meta-Llama-3.1-405B and released by Nous Research. It introduces a hybrid reasoning mode, where the model can choose to deliberate internally with...
Reasoning StructuredContext 131K Output ≤ 117K$1.00/1M $3.00/1M out cache $0.20/1MStable Image Understanding -
#415
DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...
Tools Reasoning StructuredContext 163K Output ≤ 147K$0.25/1M $0.95/1M out cache $0.13/1MStable Image Understanding -
#416
gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...
Tools Reasoning StructuredContext 131K Output ≤ 117K$0.02/1M $0.08/1M out cache $0.01/1MStable Image Understanding -
#417
gpt-oss-20b is an open-weight 21B parameter model released by OpenAI under the Apache 2.0 license. It uses a Mixture-of-Experts (MoE) architecture with 3.6B active parameters per forward pass, optimized for...
Tools Reasoning StructuredContext 131K Output ≤ 117K$0.02/1M $0.11/1M outStable Image Understanding -
#418
Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation. [Blog Post](https://mistral.ai/news/codestral-25-08)
Tools Docs StructuredContext 256K Output ≤ 204K$0.30/1M $0.90/1M out cache $0.03/1MStable Image Understanding -
#419
Qwen3-Coder-30B-A3B-Instruct is a 30.5B parameter Mixture-of-Experts (MoE) model with 128 experts (8 active per forward pass), designed for advanced code generation, repository-scale understanding, and agentic tool use. Built on the...
Tools StructuredContext 262K Output ≤ 235K$0.15/1M $0.60/1M outStable Image Understanding -
#420
Qwen3-30B-A3B-Instruct-2507 is a 30.5B-parameter mixture-of-experts language model from Qwen, with 3.3B active parameters per inference. It operates in non-thinking mode and is designed for high-quality instruction following, multilingual understanding, and...
Tools StructuredContext 262K Output ≤ 235K$0.10/1M $0.30/1M out cache $0.10/1MStable Image Understanding