Benchmark Almanac

Browser Use LinkedIn

opencode + Claude Haiku 5.5 98.7 tops its leaderboard Claude Haiku 5.5 97.9 leads as a model on its own
Not in model cards

Agents answer plain-language questions about LinkedIn using recorded page snapshots. The leaderboard reports correctness, cost, and time per task.

On November 19, 2026, Claude Haiku 5.5 led with a score of 98.7 and 100% correctness. DeepSeek V4.1 Flash followed at 98.6.

Made by
lowgular
Released
November 2026
Kind
Web browsing
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Screens → Actions, Text
Interaction
Agentic
Environment
Browser, LinkedIn
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
In views
Computer use, Agents
Strongest model tested
Claude Opus 5.5 #5 on the capability index
Last model added
November 2026
Scores refreshed
November 20th, 2026
21
Entries on its own leaderboard, 9 models we track

opencode + Claude Haiku 5.5 tops the leaderboard with 98.7 Claude Haiku 5.5 leads with 97.9

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran on its own with several settings, we keep its best. Every run is in the next chart. Agents other companies built around a model get a pale shade of that model's lab color. Gray marks blends and entries not in our model list, such as fine-tunes and older models. None of these count toward leaders. The best model on its own is Claude Haiku 5.5, at 97.9. Agents built around a model, such as opencode + Claude Haiku 5.5, are their own entries. We don't count them as the model's score. This view shows only models we track, each at its best.

Source: Official Browser Use LinkedIn leaderboard

#EntryScoreSettingReleasedRun onSource
1 opencode + Claude Haiku 5.5 98.7 linkedin-po@0.1.0 Oct 2026 Nov 2026 Official leaderboard
2 opencode + DeepSeek V4.1 Flash 98.6 linkedin-po@0.1.0 Sep 2026 Nov 2026 Official leaderboard
3 opencode + Qwen3.8 Flash 98.0 linkedin-po@0.1.0 Aug 2026 Nov 2026 Official leaderboard
4 Claude Haiku 5.5 97.9 linkedin-po@0.1.0, claude-code Oct 2026 Nov 2026 Official leaderboard
5 opencode + GLM-5.3-Flash 97.5 linkedin-po@0.1.0 Aug 2026 Nov 2026 Official leaderboard
6 opencode + DeepSeek-V4-Pro 94.7 linkedin-po@0.1.0 Apr 2026 Nov 2026 Official leaderboard
7 GPT-6.1 Terra 91.9 linkedin-po@0.1.0, codex Oct 2026 Nov 2026 Official leaderboard
8 opencode + GPT-6 Luna 90.7 linkedin-po@0.1.0 Sep 2026 Nov 2026 Official leaderboard
9 GPT-6 Luna 90.5 linkedin-po@0.1.0, codex Sep 2026 Nov 2026 Official leaderboard
10 opencode + Gemini 3.8 Flash 86.2 linkedin-po@0.1.0 Sep 2026 Nov 2026 Official leaderboard
11 Gemini 4 Hydrogen 86.2 linkedin-po@0.1.0, antigravity-cli Nov 2026 Nov 2026 Official leaderboard
12 GPT-6.1 Sol 84.9 linkedin-po@0.1.0, codex Sep 2026 Nov 2026 Official leaderboard
13 Gemini 3.8 Flash 81.0 linkedin-po@0.1.0, gemini-cli Sep 2026 Nov 2026 Official leaderboard
14 opencode + GLM-5.3 77.3 linkedin-po@0.1.0 Aug 2026 Nov 2026 Official leaderboard
15 opencode + Tencent Hy3 75.2 linkedin-po@0.1.0 Jul 2026 Nov 2026 Official leaderboard
16 Claude Sonnet 5.5 68.5 linkedin-po@0.1.0, claude-code Sep 2026 Nov 2026 Official leaderboard
17 opencode + Gemini 3.5 Flash-Lite 66.2 linkedin-po@0.1.0 Jul 2026 Nov 2026 Official leaderboard
18 Claude Opus 5.5 65.5 linkedin-po@0.1.0, claude-code Sep 2026 Nov 2026 Official leaderboard
19 opencode + MiMo-V2.6 Flash 44.5 linkedin-po@0.1.0 Sep 2026 Nov 2026 Official leaderboard
20 opencode + Qwen 3.8 Max 40.9 linkedin-po@0.1.0 Aug 2026 Nov 2026 Official leaderboard
21 Gemini 3.5 Flash-Lite 32.1 linkedin-po@0.1.0, gemini-cli Jul 2026 Nov 2026 Official leaderboard
#ModelScoreSettingReleasedRun onSource
1 Claude Haiku 5.5 97.9 linkedin-po@0.1.0, claude-code Oct 2026 Nov 2026 Official leaderboard
2 GPT-6.1 Terra 91.9 linkedin-po@0.1.0, codex Oct 2026 Nov 2026 Official leaderboard
3 GPT-6 Luna 90.5 linkedin-po@0.1.0, codex Sep 2026 Nov 2026 Official leaderboard
4 Gemini 4 Hydrogen 86.2 linkedin-po@0.1.0, antigravity-cli Nov 2026 Nov 2026 Official leaderboard
5 GPT-6.1 Sol 84.9 linkedin-po@0.1.0, codex Sep 2026 Nov 2026 Official leaderboard
6 Gemini 3.8 Flash 81.0 linkedin-po@0.1.0, gemini-cli Sep 2026 Nov 2026 Official leaderboard
7 Claude Sonnet 5.5 68.5 linkedin-po@0.1.0, claude-code Sep 2026 Nov 2026 Official leaderboard
8 Claude Opus 5.5 65.5 linkedin-po@0.1.0, claude-code Sep 2026 Nov 2026 Official leaderboard
9 Gemini 3.5 Flash-Lite 32.1 linkedin-po@0.1.0, gemini-cli Jul 2026 Nov 2026 Official leaderboard

The best score, 98.7, costs $0.006 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Agents other companies built, in a pale shade of the model’s lab Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 80.0 100.0 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale opencode + Claude Haiku 5.5, linkedin-po@0.1.0 opencode + GPT-6 Luna, linkedin-po@0.1.0 opencode + MiMo-V2.6 Flash, linkedin-po@0.1.0 0.00 20.0 40.0 60.0 80.0 100.0 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale opencode + Claude Haiku 5.5, linkedin-po@0.1.0 opencode + GPT-6 Luna, linkedin-po@0.1.0

Source: Official Browser Use LinkedIn leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
opencode + Claude Haiku 5.5 linkedin-po@0.1.0 opencode 98.7 $0.006
opencode + DeepSeek V4.1 Flash linkedin-po@0.1.0 opencode 98.6 $0.007
opencode + Qwen3.8 Flash linkedin-po@0.1.0 opencode 98.0 $0.01
Claude Haiku 5.5 linkedin-po@0.1.0 claude-code 97.9 $0.01
opencode + GLM-5.3-Flash linkedin-po@0.1.0 opencode 97.5 $0.01
opencode + DeepSeek-V4-Pro linkedin-po@0.1.0 opencode 94.7 $0.03
GPT-6.1 Terra linkedin-po@0.1.0 codex 91.9 $0.04
opencode + GPT-6 Luna linkedin-po@0.1.0 opencode 90.7 $0.006
GPT-6 Luna linkedin-po@0.1.0 codex 90.5 $0.007
Gemini 4 Hydrogen linkedin-po@0.1.0 antigravity-cli 86.2 $0.08
opencode + Gemini 3.8 Flash linkedin-po@0.1.0 opencode 86.2 $0.08
GPT-6.1 Sol linkedin-po@0.1.0 codex 84.9 $0.09
Gemini 3.8 Flash linkedin-po@0.1.0 gemini-cli 81.0 $0.12
opencode + GLM-5.3 linkedin-po@0.1.0 opencode 77.3 $0.15
opencode + Tencent Hy3 linkedin-po@0.1.0 opencode 75.2 $0.01
Claude Sonnet 5.5 linkedin-po@0.1.0 claude-code 68.5 $0.11
opencode + Gemini 3.5 Flash-Lite linkedin-po@0.1.0 opencode 66.2 $0.04
Claude Opus 5.5 linkedin-po@0.1.0 claude-code 65.5 $0.26
opencode + MiMo-V2.6 Flash linkedin-po@0.1.0 opencode 44.5 $0.005
opencode + Qwen 3.8 Max linkedin-po@0.1.0 opencode 40.9 $0.15
Gemini 3.5 Flash-Lite linkedin-po@0.1.0 gemini-cli 32.1 $0.04

The best score, 97.9, costs $0.01 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

OpenAI Anthropic Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 80.0 100.0 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Claude Haiku 5.5, linkedin-po@0.1.0 GPT-6 Luna, linkedin-po@0.1.0 0.00 20.0 40.0 60.0 80.0 100.0 $0.001 $0.01 $0.10 $1.00 Cost per task, log scale Claude Haiku 5.5, linkedin-po@0.1.0 GPT-6 Luna, linkedin-po@0.1.0

Source: Official Browser Use LinkedIn leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Haiku 5.5 linkedin-po@0.1.0 claude-code 97.9 $0.01
GPT-6.1 Terra linkedin-po@0.1.0 codex 91.9 $0.04
GPT-6 Luna linkedin-po@0.1.0 codex 90.5 $0.007
Gemini 4 Hydrogen linkedin-po@0.1.0 antigravity-cli 86.2 $0.08
GPT-6.1 Sol linkedin-po@0.1.0 codex 84.9 $0.09
Gemini 3.8 Flash linkedin-po@0.1.0 gemini-cli 81.0 $0.12
Claude Sonnet 5.5 linkedin-po@0.1.0 claude-code 68.5 $0.11
Claude Opus 5.5 linkedin-po@0.1.0 claude-code 65.5 $0.26
Gemini 3.5 Flash-Lite linkedin-po@0.1.0 gemini-cli 32.1 $0.04

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard 1 leader

Source: Official Browser Use LinkedIn leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
20.0 40.0 60.0 80.0 100.0 2027 20.0 40.0 60.0 80.0 100.0 2027

Source: Official Browser Use LinkedIn leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Computer and browser use

All 119 in Computer and browser use
  • BrowseComp In use Hard-to-find facts that need persistent web browsing. Short verifiable answers graded for exact correctness. Claude Fable 5 93.3%
  • ScreenSpot-Pro In use GUI grounding on high-resolution professional-software screenshots. The model must locate the target UI element. GPT-6 Astra 92.7%
  • AndroidWorld In use Agents complete 116 tasks across 20 real Android apps in an emulator. Each task checks the device's system state to judge success. GPT-6 Astra 100%
  • OSWorld 2.1 In use Agents complete 108 long-horizon workflows across 31 self-hosted websites and desktop apps, averaging 318 tool calls. Scored by binary completion and checkpoint partial credit. Claude Opus 5 44.3%
  • WebArena-Verified In use Agents complete 812 audited WebArena tasks on self-hosted websites. Deterministic evaluators check the agent's answer and captured network traces. Muse Spark 1.1 69%
  • OSWorld 2.1 (offline subset) In use Agents complete the 82 OSWorld 2.1 tasks on the official offline list, with no internet access. Scored by checkpoint partial credit and strict pass rate. Claude Fable 5.2 89.3%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (21 entries).
  • Pass Models scored 9 models we track, plus 12 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Gemini 4 Hydrogen.
  • Pass Frontier models 4 of the top 10 have a score, led by Claude Opus 5.5 (#5).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on hub.harborframework.com: 21 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check