Decision Intelligence Platform
What AI should you use for this job?
Tell us your goal and constraints. Get a sourced recommendation with verification dates, trade-offs, and the alternatives we rejected — so you can check our reasoning.
A real determination, with its sources
What a recommendation looks like
Steps, tools, and cost come straight from the catalog — with every claim linked and dated. Rejected options are shown struck-through, with the reason.
Goal · Construct a complete, multi-page SaaS marketing website featuring pricing tiers, feature lists, documentation pages, and contact captures.
- 01 · Site Builder
Generates multi-page SaaS websites including pricing, features, and about grids; Pro starts at $25/month for 100 monthly credits.
- 02 · Hero Art
Renders the custom hero backgrounds and product mockups; Basic starts around $10/month with no free plan.
StackPaid Site: Lovable + Midjourney
Estimated cost~$35/month (Lovable Pro $25 + Midjourney Basic $10)
Lovable builds and hosts the multi-page marketing site from prompts, and Midjourney supplies the custom hero art and mockups that templates can't.
Open the full workflowWorkflows
Documented pipelines
Three-node stacks pulled from the catalog. Every step links to a sourced tool.
Editor's picks
Recently verified
Each listing carries its real verification state — a linked source and a check date, or an honest 'reference entry' label.
ChatGPT
VERIFIED AUG 16Industry-standard chatbot by OpenAI supporting prose writing, code compilation, and reasoning models.
Claude
VERIFIED AUG 14Advanced language model by Anthropic noted for complex programming, empathy writing, and Artifacts.
Cursor
VERIFIED SEP 10AI-first editor with a full cloud-agent platform: autonomous coding, self-hosted execution, and code hosting.
Midjourney
VERIFIED AUG 14Artistic text-to-image synthesizer noted for photorealistic composition lighting and consistency.
Changelog
Latest updates
Recent releases and capability expansions, stamped by date.
Honestack Benchmark Lab
Evidence in Models HubMeasurement, not marketing
Every bar is one cited benchmark result with a source URL and verification date. We never mix benchmarks into a universal score, and a missing number is shown as a gap — never filled in.
Evidence records
90
sourced benchmark results
Models with evidence
28/31
from the Models Hub
Benchmarks
13
each with a real source
Sourced tools
222/222
222 verified today
Free / freemium
187
of catalog
Model evidence leaderboard
Per-benchmark results · not a universal score
LMArena (Elo)
Claude Fable 5
Anthropic · text · verified Sep 24, 2026
1506
LMArena (Elo)
Claude Fable 5.1
Anthropic · text · verified Sep 24, 2026
1498
LMArena (Elo)
Claude Opus 5
Anthropic · text · verified Sep 24, 2026
1493
LMArena (Elo)
Gemini 3.8 Flash
Google · text · verified Sep 24, 2026
1493
LMArena (Elo)
Gemini 3.7 Flash
Google · text · verified Sep 24, 2026
1490
LMArena (Elo)
Gemini 3.1 Pro
Google DeepMind · text · verified Sep 24, 2026
1487
LMArena (Elo)
Kimi K3
Moonshot AI · text · verified Sep 24, 2026
1485
LMArena (Elo)
GPT-5.6 Sol
OpenAI · text · verified Sep 24, 2026
1484
LMArena (Elo)
GLM-5.3
Z.ai · text · verified Sep 24, 2026
1483
LMArena (Elo)
Qwen3.8 Max
Alibaba · text · verified Sep 24, 2026
1481
LMArena (Elo)
GPT-6 Astra
OpenAI · text · verified Sep 24, 2026
1480
LMArena (Elo)
DeepSeek V4 Pro
DeepSeek · text · verified Sep 24, 2026
1457
SWE-bench Verified
Claude Opus 5
Anthropic · Verified · verified Sep 24, 2026
96.0%
GPQA
GPT-6 Astra
OpenAI · Diamond · verified Sep 24, 2026
96.0%
SWE-bench Verified
Claude Fable 5
Anthropic · Verified · verified Aug 16, 2026
95.5%
SWE-bench Verified
Claude Sonnet 5
Anthropic · Verified · verified Aug 16, 2026
95.5%
GPQA
GPT-5.6 Sol
OpenAI · Diamond · verified Sep 24, 2026
94.6%
GPQA
Gemini 3.1 Pro
Google DeepMind · Diamond · verified Sep 24, 2026
94.3%
GPQA
Kimi K3
Moonshot AI · Diamond · verified Sep 10, 2026
93.5%
GPQA
GPT-5.6 Terra
OpenAI · Diamond · verified Sep 24, 2026
92.9%
GPQA
Qwen3.8 Max
Alibaba · Diamond · verified Sep 10, 2026
92.6%
GPQA
GPT-5.6 Luna
OpenAI · Diamond · verified Sep 24, 2026
92.3%
Terminal-Bench
Kimi K3
Moonshot AI · v2.1 · verified Sep 10, 2026
88.3%
Terminal-Bench
GLM-5.3
Z.ai · v2.1 · verified Sep 10, 2026
88.2%
Terminal-Bench
DeepSeek V4 Pro
DeepSeek · v2.1 · verified Sep 10, 2026
87.9%
Terminal-Bench
DeepSeek V4 Flash
DeepSeek · v2.1 · verified Sep 10, 2026
82.7%
SWE-bench Verified
Gemini 3.1 Pro
Google DeepMind · Verified · verified Sep 24, 2026
80.6%
Artificial Analysis Coding Agent Index
GPT-5.6 Sol
OpenAI · v1.3 · verified Aug 16, 2026
80.0
SWE-bench Verified
Mistral Medium 3.5
Mistral AI · Verified · verified Aug 16, 2026
77.6%
DeepSWE
GPT-6 Astra
OpenAI · v1.1 · verified Sep 24, 2026
74.1%
Terminal-Bench
Qwen3.8-27B
Alibaba · v2.1 · verified Sep 10, 2026
73.0%
DeepSWE
GPT-5.6 Sol
OpenAI · v1.1 · verified Sep 24, 2026
72.7%
Artificial Analysis Coding Index
Claude Sonnet 5
Anthropic · current · verified Aug 16, 2026
71.3
Humanity's Last Exam
Claude Opus 5.5
Anthropic · with tools · verified Sep 24, 2026
67.7%
Terminal-Bench
Claude Opus 5.5
Anthropic · v4.0 · verified Sep 24, 2026
66.4%
DeepSWE
Grok 4.6
xAI · v1.1 · verified Sep 10, 2026
65.9%
Humanity's Last Exam
Claude Fable 5.1
Anthropic · with tools · verified Sep 24, 2026
65.6%
Humanity's Last Exam
Claude Opus 5
Anthropic · with tools · verified Sep 24, 2026
63.6%
DeepSWE
DeepSeek V4 Pro
DeepSeek · 113 tasks · verified Sep 10, 2026
62.7%
Terminal-Bench
Claude Opus 5.5
Anthropic · v4.0 · verified Sep 24, 2026
59.6%
Artificial Analysis Coding Index
Qwen3-Coder-Next
Alibaba · current · verified Aug 16, 2026
58.2
Artificial Analysis Intelligence Index
Claude Opus 5.5
Anthropic · v4.3 · verified Sep 24, 2026
58
Terminal-Bench
GPT-6 Astra
OpenAI · v4.0 · verified Sep 24, 2026
57.9%
Humanity's Last Exam
GPT-6 Astra
OpenAI · with tools · verified Sep 24, 2026
57.2%
Artificial Analysis Coding Agent Index
GPT-6 Sol
OpenAI · Codex harness · verified Sep 24, 2026
57
Terminal-Bench
Claude Fable 5.1
Anthropic · v4.0 · verified Sep 24, 2026
55.8%
Humanity's Last Exam
Gemini 3.8 Flash
Google · HLE-Verified · verified Sep 10, 2026
54.9%
DeepSWE
DeepSeek V4 Flash
DeepSeek · 113 tasks · verified Sep 10, 2026
54.4%
Artificial Analysis Intelligence Index
Claude Fable 5.1
Anthropic · v4.3 · verified Sep 24, 2026
53
Artificial Analysis Intelligence Index
GPT-6 Astra
OpenAI · v4.3 · verified Sep 24, 2026
53
Terminal-Bench
Claude Opus 5
Anthropic · v4.0 · verified Sep 24, 2026
51.8%
Artificial Analysis Intelligence Index
Claude Opus 5
Anthropic · v4.3 · verified Sep 24, 2026
51
Artificial Analysis Intelligence Index
Claude Fable 5
Anthropic · v4.3 · verified Sep 24, 2026
50
Artificial Analysis Intelligence Index
GPT-6 Sol
OpenAI · v4.3 · verified Sep 24, 2026
48
Humanity's Last Exam
GPT-6 Sol
OpenAI · AA run · verified Sep 24, 2026
48%
Artificial Analysis Intelligence Index
GPT-5.6 Sol
OpenAI · v4.3 · verified Sep 24, 2026
47
Artificial Analysis Intelligence Index
Qwen3.8 Max
Alibaba · v4.3 · verified Sep 24, 2026
45
Artificial Analysis Intelligence Index
GLM-5.3
Z.ai · v4.3 · verified Sep 24, 2026
45
Terminal-Bench
Claude Fable 5
Anthropic · v4.0 · verified Sep 24, 2026
44.5%
Terminal-Bench
GPT-6 Sol
OpenAI · v4.0 · verified Sep 24, 2026
44%
Artificial Analysis Intelligence Index
Grok 4.6
xAI · v4.3 · verified Sep 24, 2026
44
Artificial Analysis Intelligence Index
Kimi K3
Moonshot AI · v4.3 · verified Sep 24, 2026
44
Artificial Analysis Intelligence Index
GPT-5.6 Terra
OpenAI · v4.3 · verified Sep 24, 2026
42
Terminal-Bench
GLM-5.3
Z.ai · v4.0 · verified Sep 24, 2026
41.8%
Artificial Analysis Coding Agent Index
GPT-6 Luna
OpenAI · Codex harness · verified Sep 24, 2026
41
Artificial Analysis Intelligence Index
Gemini 3.8 Flash
Google · v4.3 · verified Sep 24, 2026
41
Artificial Analysis Intelligence Index
Gemini 3.7 Flash
Google · v4.3 · verified Sep 24, 2026
40
Artificial Analysis Intelligence Index
GLM-5.2
Z.ai · v4.3 · verified Sep 24, 2026
39
Artificial Analysis Intelligence Index
DeepSeek V4 Flash
DeepSeek · v4.3 · verified Sep 24, 2026
39
Artificial Analysis Intelligence Index
Claude Sonnet 5
Anthropic · v4.3 · verified Sep 24, 2026
38
Terminal-Bench
GPT-5.6 Sol
OpenAI · v4.0 · verified Sep 24, 2026
37.3%
Artificial Analysis Intelligence Index
GPT-6 Luna
OpenAI · v4.3 · verified Sep 24, 2026
37
Artificial Analysis Intelligence Index
GPT-5.6 Luna
OpenAI · v4.3 · verified Sep 24, 2026
37
Artificial Analysis Intelligence Index
DeepSeek V4 Pro
DeepSeek · v4.3 · verified Sep 24, 2026
36
Artificial Analysis Intelligence Index
Qwen3.8-27B
Alibaba · v4.3 · verified Sep 24, 2026
34
Artificial Analysis Intelligence Index
Gemini 3.1 Pro
Google DeepMind · v4.3 · verified Sep 24, 2026
30
Artificial Analysis Intelligence Index
MiniMax M3
MiniMax · v4.3 · verified Sep 24, 2026
29
Terminal-Bench
GPT-5.6 Terra
OpenAI · v4.0 · verified Sep 24, 2026
21.5%
Terminal-Bench
Grok 4.6
xAI · v4.0 · verified Sep 24, 2026
20.3%
Artificial Analysis Intelligence Index
K-EXAONE 2.0
LG AI Research · v4.3 · verified Sep 24, 2026
20
Terminal-Bench
Gemini 3.8 Flash
Google · v4.0 · verified Sep 24, 2026
19.1%
Artificial Analysis Intelligence Index
Gemma 4
Google · v4.3 · verified Sep 24, 2026
19
Terminal-Bench
GPT-5.6 Luna
OpenAI · v4.0 · verified Sep 24, 2026
17.3%
Artificial Analysis Intelligence Index
Mistral Medium 3.5
Mistral AI · v4.3 · verified Sep 24, 2026
14
Terminal-Bench
GPT-6 Luna
OpenAI · v4.0 · verified Sep 24, 2026
13%
Terminal-Bench
Kimi K3
Moonshot AI · v4.0 · verified Sep 24, 2026
12.6%
Terminal-Bench
Claude Sonnet 5
Anthropic · v4.0 · verified Sep 24, 2026
12.4%
Terminal-Bench
Gemini 3.7 Flash
Google · v4.0 · verified Sep 24, 2026
11.2%
Artificial Analysis Intelligence Index
Llama 4 Maverick
Meta · v4.3 · verified Sep 24, 2026
10
Artificial Analysis Intelligence Index
K-EXAONE 2.0
LG AI Research · current · verified Sep 10, 2026
listed
ProprietaryOpen WeightsOpen Source90 sourced results
How the Lab scores
What counts
Cited benchmark results from the Models Hub — each with a source URL, benchmark version, and verification date.
What doesn't
No public ratings, no crowd-sourced scores, no fabricated numbers. Benchmarks are never averaged into one number.
Last check
Each evidence record carries its own verification date. Freshness degrades automatically when nothing is re-verified.
Evidence table
90 records · filter above| Version | Type | Source | ||||
|---|---|---|---|---|---|---|
| 1506 | Claude Fable 5 | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1498 | Claude Fable 5.1 | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1493 | Claude Opus 5 | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1493 | Gemini 3.8 Flash | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1490 | Gemini 3.7 Flash | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1487 | Gemini 3.1 Pro | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1485 | Kimi K3 | LMArena (Elo) | text | Open Weights | Sep 24, 2026 | link |
| 1484 | GPT-5.6 Sol | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1483 | GLM-5.3 | LMArena (Elo) | text | Open Weights | Sep 24, 2026 | link |
| 1481 | Qwen3.8 Max | LMArena (Elo) | text | Open Weights | Sep 24, 2026 | link |
| 1480 | GPT-6 Astra | LMArena (Elo) | text | Proprietary | Sep 24, 2026 | link |
| 1457 | DeepSeek V4 Pro | LMArena (Elo) | text | Open Weights | Sep 24, 2026 | link |
| 96.0% | Claude Opus 5 | SWE-bench Verified | Verified | Proprietary | Sep 24, 2026 | link |
| 96.0% | GPT-6 Astra | GPQA | Diamond | Proprietary | Sep 24, 2026 | link |
| 95.5% | Claude Fable 5 | SWE-bench Verified | Verified | Proprietary | Aug 16, 2026 | link |
| 95.5% | Claude Sonnet 5 | SWE-bench Verified | Verified | Proprietary | Aug 16, 2026 | link |
| 94.6% | GPT-5.6 Sol | GPQA | Diamond | Proprietary | Sep 24, 2026 | link |
| 94.3% | Gemini 3.1 Pro | GPQA | Diamond | Proprietary | Sep 24, 2026 | link |
| 93.5% | Kimi K3 | GPQA | Diamond | Open Weights | Sep 10, 2026 | link |
| 92.9% | GPT-5.6 Terra | GPQA | Diamond | Proprietary | Sep 24, 2026 | link |
| 92.6% | Qwen3.8 Max | GPQA | Diamond | Open Weights | Sep 10, 2026 | link |
| 92.3% | GPT-5.6 Luna | GPQA | Diamond | Proprietary | Sep 24, 2026 | link |
| 88.3% | Kimi K3 | Terminal-Bench | v2.1 | Open Weights | Sep 10, 2026 | link |
| 88.2% | GLM-5.3 | Terminal-Bench | v2.1 | Open Weights | Sep 10, 2026 | link |
| 87.9% | DeepSeek V4 Pro | Terminal-Bench | v2.1 | Open Weights | Sep 10, 2026 | link |
| 82.7% | DeepSeek V4 Flash | Terminal-Bench | v2.1 | Open Weights | Sep 10, 2026 | link |
| 80.6% | Gemini 3.1 Pro | SWE-bench Verified | Verified | Proprietary | Sep 24, 2026 | link |
| 80.0 | GPT-5.6 Sol | Artificial Analysis Coding Agent Index | v1.3 | Proprietary | Aug 16, 2026 | link |
| 77.6% | Mistral Medium 3.5 | SWE-bench Verified | Verified | Open Weights | Aug 16, 2026 | link |
| 74.1% | GPT-6 Astra | DeepSWE | v1.1 | Proprietary | Sep 24, 2026 | link |
Showing top 30 of 90 records — full table lives in the Models Hub.
Head-to-head
Comparisons
Structured spec breakdowns and objective verdict outcomes.
A comparison of general conversational power: ChatGPT’s multi-tool versatility versus Claude’s precise prose and technical coding abilities.
Choose Claude if you seek precise coding help, front-end visual prototyping, or natural prose writing. Choose ChatGPT if you rely on built-in image generators, voice-driven interfaces, or custom GPT bot lists.
Head-to-head review of conversational structures: ChatGPT’s specialized reasoning models versus Gemini’s ultra-massive context window and Google Workspace connections.
Choose Gemini if you require extensive file ingestion (such as summarizing entire books or hour-long zoom calls) and rely heavily on Google Workspace. Choose ChatGPT if you need top-tier mathematical reasoning logic or customized coding help.
The developer's dilemma: Cursor's multi-file agentic code editing versus GitHub Copilot's seamless, industry-standard editor autocomplete.
Choose Cursor if you want an autonomous assistant to build entire features, generate new folders, and handle server terminals. Choose GitHub Copilot if you want high-speed, secure, inline autocomplete inside IntelliJ, PyCharm, or stock VS Code.
The Honestack rule
Source-led listings. Every claim links to a source — or says it doesn't have one yet.
No public ratings, no crowd-sourced scoring, no fabricated benchmarks. Each listing links to an official source and records when its details were last checked. Listings without sources are labeled honestly as reference entries.
Read the methodology- Tools on file
- 222
- Source-linked
- 222/222
- Documented workflows
- 94