We compare the major LLM APIs as of May 2026 — Google Gemini 2.0 / OpenAI GPT-4o & o1 / Anthropic Claude Opus 4 & Sonnet 4 / Meta Llama 3.3. Code generation, reasoning, multimodal, pricing, and how to use them through each cloud — it's all here.
| Model | Context | Input ($/M tok) | Output ($/M tok) | Multimodal |
|---|---|---|---|---|
| Gemini 2.0 Flash | 1M | $0.075 | $0.30 | Text + image + audio + video |
| Gemini 2.0 Pro | 2M | $1.25 | $5 | Text + image + audio + video (2h) |
| GPT-4o mini | 128k | $0.15 | $0.60 | Text + image |
| GPT-4o | 128k | $2.50 | $10 | Text + image + audio |
| OpenAI o1 | 200k | $15 | $60 | Text |
| Claude Haiku 4 | 200k | $0.80 | $4 | Text + image |
| Claude Sonnet 4 | 200k | $3 | $15 | Text + image |
| Claude Opus 4 | 1M (extended) | $15 | $75 | Text + image |
| Llama 3.3 70B | 128k | OSS (endpoint cost) | Same as input | Text |
| Benchmark | Gemini 2.0 Pro | GPT-4o | Claude Sonnet 4 | Llama 3.3 70B |
|---|---|---|---|---|
| MMLU (general knowledge) | ~85 | ~88 | ~89 | ~86 |
| HumanEval (code) | ~85 | ~90 | ~95 (Sonnet 4) | ~80 |
| SWE-bench (real-world code fixes) | ~50 | ~55 | ~70 (Opus 4) | ~30 |
| MATH | ~75 | ~80 (o1: 95) | ~85 | ~70 |
| Multimodal (image) | ~85 | ~85 | ~80 | — |
| Model | Vertex AI | AWS Bedrock | Azure OpenAI / AI Foundry |
|---|---|---|---|
| Gemini | ◎ | — | — |
| GPT-4o / o1 | — | — | ◎ |
| Claude | ◎ | ◎ | — |
| Llama | ◎ | ◎ | ○ |
| Mistral | ◎ | ◎ | ◎ |
| Use case | Recommended | Why |
|---|---|---|
| High-volume chatbots (low cost) | Gemini Flash | Cheapest at $0.075/M tok |
| Coding agents | Claude Sonnet 4 / Opus 4 | Industry-leading SWE-bench score |
| Math and complex reasoning | OpenAI o1 | Built-in Chain-of-Thought |
| Multimodal video analysis | Gemini 2.0 Pro | 2-hour video in a single request |
| Voice interaction | GPT-4o | Real-time voice |
| RAG (large document sets) | Gemini Pro 2M tok | Long context window |
| Self-hosting | Llama 3.3 / Gemma 2 | OSS |
| Model | Price | Dimensions | Multilingual |
|---|---|---|---|
| OpenAI text-embedding-3-small | $0.02/M tok | 1536 | ◎ |
| OpenAI text-embedding-3-large | $0.13/M tok | 3072 | ◎ |
| Google text-embedding-005 | $0.025/1000 chars | 768 | ◎ |
| Cohere embed-multilingual-v3 | $0.10/M tok | 1024 | ◎ |
Which should I choose: Gemini, GPT, or Claude?
Cost-focused → Gemini Flash; code generation → Claude Sonnet 4; complex reasoning → GPT-4o / o1 / Claude Opus 4; multimodal video → Gemini Pro.
What stands out about Claude Opus 4 (1M context)?
Released 2025-09 with industry-leading coding ability, reasoning accuracy, and a 1M-token context window. You can fit an entire large codebase into a single request.
What are the strengths of Gemini 2.0 Pro?
Up to a 2M-token context window, a single request can process 2 hours of video, Google Search Grounding, and a low price ($1.25/M tok). Best-in-class multimodal.
What are the strengths of GPT-4o / o1?
GPT-4o supports real-time voice and video; o1 uses Chain-of-Thought reasoning for strong math and code; the largest ecosystem thanks to ChatGPT's reach.
How do prices compare?
Flagship comparison (input/output $/M tok): Gemini Pro $1.25/$5, GPT-4o $2.50/$10, Claude Sonnet 4 $3/$15, Claude Opus 4 $15/$75.
Can I use these via GCP / AWS / Azure?
Vertex AI = Gemini + Claude (official Anthropic); Bedrock = Claude + Llama + Mistral; Azure OpenAI = GPT-4o / o1 only. For multi-cloud, Vertex + Azure OpenAI is the recommended combo.
What about open models?
Llama 3.3 (Meta), Mistral Large, Gemma 2 (Google), DeepSeek R1 (reasoning-focused), and Qwen 2.5 (Alibaba). Self-host and fine-tune freely.
Which embedding model should I use for RAG?
OpenAI text-embedding-3-small ($0.02/M tok), Google text-embedding-005, and Cohere embed-multilingual-v3. Google / Cohere lead on multilingual quality; OpenAI wins on cost.
Related articles: LLM comparisons
GCP vs Azure 完全比較｜Compute・Storage・AI・料金・認定 (2026)
Google Cloud と Microsoft Azure を徹底比較。Compute Engine vs Azure VM、BigQuery vs Synapse / Fabric、Cloud Run vs Container Apps、Gemini vs Azure OpenAI、ID 統合、ハイブリッド、認定試験、料金を 2026 年最新版で網羅。
GCP vs AWS ストレージ・DB 徹底比較｜GCS/S3・BigQuery/Redshift・Spanner/DynamoDB (2026)
GCP と AWS のストレージ・データベースを徹底比較。Cloud Storage vs S3、BigQuery vs Redshift、Spanner vs DynamoDB / Aurora DSQL、Cloud SQL vs RDS、AlloyDB vs Aurora、Firestore vs DynamoDB、Bigtable vs DynamoDB を 2026 年最新版で網羅。
Gemini API 完全ガイド｜料金・モデル選定・マルチモーダル・Function Calling (2026)
Google Gemini API の全機能解説。Gemini 2.0 Flash / Pro / Ultra のモデル選定、料金 ($0.075/M tok〜)、マルチモーダル、Function Calling、Context Caching、Safety Filter、Google AI Studio vs Vertex AI 使い分けを 2026 年最新版で網羅。
Vertex AI vs SageMaker vs Azure ML 徹底比較｜MLOps プラットフォーム選び方 (2026)
Google Vertex AI / AWS SageMaker / Azure ML の徹底比較。Gen AI 統合 (Gemini / Bedrock / Azure OpenAI)、AutoML、Pipelines、Feature Store、GPU/TPU、料金、認定試験を 2026 年最新版で網羅。
* Each product is owned by its respective trademark holder (Google / OpenAI / Anthropic / Meta). Check each vendor's official site for the latest specs and pricing. Benchmark scores reflect typical evaluations at publication time and vary by use case.
Practice with certification-focused question sets
Visit the GCP exam prep pageNicheeLab Editorial Team
NicheeLab editorial team focused on data engineering and cloud certification learning. Content is structured around practical study needs and official exam domains.
Google Cloud Certification Roadmap (2026)
Choose your GCP certification path — Foundational, Associate...
CDL Cloud Digital Leader: Complete Exam Guide (2026)
Pass the Cloud Digital Leader exam — cloud business value, G...
GAIL Generative AI Leader: Complete Exam Guide (2026)
Pass the Generative AI Leader exam — Gemini, Vertex AI, Work...
Vertex AI Fundamentals for GCP Certs (2026)
Vertex AI basics every cert candidate needs — Workbench, Pip...
Associate Cloud Engineer (ACE): Complete Guide (2026)
Pass the Associate Cloud Engineer exam — Console, gcloud, pr...