RANK #33 IN CODING AGENTS
Poolside
verified v2026.9 Coverage TierOpen-weight code models (Laguna) with self-serve API
VENDOR: Poolside
•
LICENSE: Proprietary / SaaS
•
SECURITY: Standard cloud security
•
LAST BENCHMARK: Verified 2026-09-15
Empirical Performance Audit
Macro Intelligence Index
80.4
/100
Coverage Tier
Relative Velocity
#33 of 55
80.4/100 composite
Archive record
4 recorded features
Source: poolside.ai
Output quality
score.dims.quality
source: archive
Output Quality & Reliability
(Weighted 25%)
84.0%
correctness and consistency of primary task output.
AiRecMark rubric v2
Feature Depth & Integrations
(Weighted 20%)
78.0%
breadth of integrations, APIs, and advanced capabilities.
AiRecMark rubric v2
Onboarding & Usability
(Weighted 15%)
76.0%
onboarding friction, UI clarity, and workflow ergonomics.
AiRecMark rubric v2
Runtime Performance & Latency
(Weighted 20%)
80.0%
latency, throughput, and stability under load.
AiRecMark rubric v2
Price-to-Value Efficiency
(Weighted 20%)
82.0%
capability delivered per pricing tier dollar.
AiRecMark rubric v2
Low-Level Verification
ISO/IEC 25010 Ref
Quantitative Technical Specification
terminal
Runtime & Core Shell
Laguna models
S (118B MoE) and XS (33B on-device).
Source: Poolside · Verified 2026-09-15
hub
Context Indexing Engine
Self-serve API
Platform keys, free for a limited time.
Source: Poolside · Verified 2026-09-15
device_hub
Protocol Specification
Pool agent
CLI plus Desktop Assistant.
Source: Poolside · Verified 2026-09-15
smart_toy
Model Agility & Routing
Open weights
Apache-2.0 licensed.
Source: Poolside · Verified 2026-09-15
Market Positioning Matrix
Open Full Matrix open_in_new
Direct Competitor Radar & Alternatives
| Tool & Version | Score | Primary Sweet Spot | Entry Pricing | Action |
|---|---|---|---|---|
|
P
Poolside
(Current)
|
80.4 | Laguna models | Free limited-time self-serve API | Selected |
|
CU
Cursor
|
94.2 | Forked VS Code Agent | $20/user/mo | Compare vs |
|
WI
Windsurf
|
91.8 | Realtime Collaborative IDE | Free / $15 Pro | Compare vs |
|
GI
GitHub Copilot
|
88.1 | Enterprise Compliance | $10/mo | Compare vs |
|
CL
Claude Code
|
86.1 | Terminal & CI/CD Pipelines | $17/mo | Compare vs |
Institutional Synthesis
Analyst Consensus Verdict
“Poolside stands out for laguna S 2.1: 118B MoE with 1M-token context. Open-weight code models (Laguna) with self-serve API anchors its proposition, and AiRecMark's five-dimension audit lands it at 80.4/100 — a pragmatic default for self-Hosting Code AI.”
— AiRecMark Senior Systems Architecture Committee
verified
Verified Strategic Strengths
- ✓ Laguna S 2.1: 118B MoE with 1M-token context
- ✓ Open weights (Apache-2.0): on Hugging Face
- ✓ OpenAI-compatible API plus: agent CLI
- ✓ Best Fit: Self-Hosting Code AI
warning
Institutional Trade-offs & Headwinds
- ! Enterprise pricing not: published
- ! Self-hosting needs serious: GPU capacity
- ! :
- ! Scope: Editorial assessment based on public information; hands-on retest pending.
Total Cost of Ownership
Transparent
Commercial Tiers
Free
$0 forever
Entry tier for evaluation and light individual use.
Entry evaluation tier
Most Deployed
Pro
Custom
Free limited-time self-serve API · open weights (Apache-2.0) on Hugging Face · enterprise pricing not published
Verified 2026-09-15 · poolside.ai
Enterprise
Custom
Dedicated infrastructure, security review, and compliance support.
SSO / SAML & audit controls
TCO MODEL NOTE: Official pricing: Free limited-time self-serve API · open weights (Apache-2.0) on Hugging Face · enterprise pricing not published. Verified via poolside.ai (2026-09-15); overages and enterprise terms bill per the official pricing page.
Archive Record
Recorded Metadata
ARCHIVE SNAPSHOT
snapshot 2026-09-15
OFFICIAL SOURCE
Source: poolside.ai
PRICING CHECKED
Snapshot 2026-09-15
SCORING MODEL
v2-5dim
RECORDED FEATURES
AiRecMark Editorial
Category Standing
Ecosystem Gravity
Category Rank
#33 of 55
Best For
Best for: Self-Hosting Code AI
Sources: Official docs, poolside.ai, community signals
N=4 recorded attributes
shield_with_heart
AIRECMARK EMPIRICAL PROTOCOL v2.4
•
Independent, un-sponsored deterministic evaluation clusters.
All benchmarks executed in sandboxed hypervisors with standardized token latency meters.